Loading…

Sign up or log in to bookmark your favorites and sync them to your phone or calendar.

Tuesday, March 12
 

9:30am PDT

OCP SONiC/SAI Pre-Summit Workshop & Hackathon - hosted by LinkedIn (Offsite) - Must Pre-Register!
Welcome to the SONiC/SAI Pre-Summit Workshop and Hackathon. During the workshop, we will have technical talks and a one-day hackathon on SONiC.
A laptop is required for hackathon. We will have 1 hour training on how to install the Virtual Switch with SONiC image on your laptop and how to use it.
Participants may team with up others. Attendees will vote for the winning team at the end of the hackathon.
Agenda
Tuesday, March 12
9:30 – 10:00 - BREAKFAST SOCIAL
10-10:10 – OPENING AND WELCOME - Sadaf Fardeen, LinkedIn
10:10 – 10:30 – KEYNOTE - Zaid Kahn, LinkedIn
10:30 – 11:00 - "SONiC OS Virtualization and Data Plane Emulation Work" - Yongfeng Liu, Alibaba
11:05 - 11:35 - "SONiC in Microsoft Azure: Enabling Mission Critical Applications" - Alberto Gonzalez Prieto and Weixi Chen, Microsoft
11:40 - 12:10 - "SONIC Extension Infrastructure" - Matty Kadosh, Mellanox
12:15 - 12:45 - "Using programmable dataplanes to transform network system high availability" - Kumaresh Perumal, Barefoot
12:45 – 1:30 – LUNCH
1:30 - 2:00 - "Developer`s Overview of Sonic", - Praveen Chaudhary, Linkdedin
2:00 - 2:30 - "Opportunities and Obstacles for Open Source Networking" - Alley Hasan and Sandeep Madhavan, Dell
2:30 - 3:00 - "Cloud-grade Routing as a Micro-service for Open Networking Platforms" - Vinay Nallamothu, Juniper
3:00 – 3:15 – BREAK
3:15 – 3:45 – "Using NBI images for SONiC system on Cisco Platforms" - Ashoka Kallappa and Jiahua Wang, Cisco
3:45 – 4:15 – "SONiC unit test and function test enhancement" - Larry Ho, Edgecore
4:15 – 4:25 – HACKATHON INTRO - Sadaf Fardeen, LinkedIn
4:25 – 4:30 – BREAK
4:30 – 5:30 – System Setup Instructions and Team up
5:30 PM - DISPERSE
Wednesday, March 13
8:00am – 4:00pm – HACKATHON (Breakfast included)
12:00 - 1:00 - LUNCH
4:00 - 5:00 - HACKATHON PRESENTATION AND AWARD
Instructions for Hackathon:
Put on your thinking caps, team up with friends, and get ready to be the geek coder!! As a part of SONiC Workshop we are organizing a Hackathon for SONiC. Here are more details of the event. Feel free to reach out with questions if any.
Rules
1. Participate alone or in teams up to a size of 4.
2. Team up with anyone in the community.
3. Free to choose any idea as long as it is useful for SONiC. Some suggested areas: test framework, debugging, performance improvement, standard interfaces, Monitoring.
4. Execute on a virtual switch or a physical switch.
5. Hackathon begins at 4PM on 12th March and ends at 4 PM on 13th March. You are free to work at home or hotel during this period and come back at 4 PM for final presentation.
6. Participants from remote/video conferencing can also join.
7. Final presentation at 4PM on 13th March includes a working demo of 7-10 min.
8. Bring your own laptop for development.
9. Development in any programming language is fine as long as it can integrate with SONiC ecosystem.
10. Use of any off the shelf tools and package is allowed as long as it has appropriate license for use with SONiC.
Instructions
1. Prepare your laptop for virtual switch based development by following instructions given here:
https://github.com/Azure/sonic-mgmt/blob/master/ansible/doc/README.testbed.VsSetup.md
2.Reach out to community and form a team before the event.
Awards
1. Three winning teams to be chosen. $1000 per team.
- Zeus Hacker: Award for most difficult project
- Athena hacker: Award for best idea
- Prometheus hacker: Award for best presentation
2. Judging committee consists of Senior Engineering leaders from SONiC community and OCP
- Haiyong Wang, Senior Director, Alibaba Group
- Lihua Yuan, Partner Developer Manager, Microsoft
- Omar Baldonado, Director of Software Engineering, Facebook
- Shawn Zandi, Director of Engineering, LinkedIn
3. Criteria for evaluation : Relevance for SONiC, complexity in execution, overall presentation and uniqueness of idea.
4. All participants will get a swag

Click here to pre-register. 


Tuesday March 12, 2019 9:30am - 5:30pm PDT
LinkedIn Building R
 
Wednesday, March 13
 

8:00am PDT

OCP SONiC/SAI Pre-Summit Workshop & Hackathon - hosted by LinkedIn (Offsite) - Must Pre-register!
Welcome to the SONiC/SAI Pre-Summit Workshop and Hackathon. During the workshop, we will have technical talks and a one-day hackathon on SONiC.
A laptop is required for hackathon. We will have 1 hour training on how to install the Virtual Switch with SONiC image on your laptop and how to use it.
Participants may team with up others. Attendees will vote for the winning team at the end of the hackathon.
Agenda
Tuesday, March 12
9:30 – 10:00 - BREAKFAST SOCIAL
10-10:10 – OPENING AND WELCOME - Sadaf Fardeen, LinkedIn
10:10 – 10:30 – KEYNOTE - Zaid Kahn, LinkedIn
10:30 – 11:00 - "SONiC OS Virtualization and Data Plane Emulation Work" - Yongfeng Liu, Alibaba
11:05 - 11:35 - "SONiC in Microsoft Azure: Enabling Mission Critical Applications" - Alberto Gonzalez Prieto and Weixi Chen, Microsoft
11:40 - 12:10 - "SONIC Extension Infrastructure" - Matty Kadosh, Mellanox
12:15 - 12:45 - "Using programmable dataplanes to transform network system high availability" - Kumaresh Perumal, Barefoot
12:45 – 1:30 – LUNCH
1:30 - 2:00 - "Developer`s Overview of Sonic", - Praveen Chaudhary, Linkdedin
2:00 - 2:30 - "Opportunities and Obstacles for Open Source Networking" - Alley Hasan and Sandeep Madhavan, Dell
2:30 - 3:00 - "Cloud-grade Routing as a Micro-service for Open Networking Platforms" - Vinay Nallamothu, Juniper
3:00 – 3:15 – BREAK
3:15 – 3:45 – "Using NBI images for SONiC system on Cisco Platforms" - Ashoka Kallappa and Jiahua Wang, Cisco
3:45 – 4:15 – "SONiC unit test and function test enhancement" - Larry Ho, Edgecore
4:15 – 4:25 – HACKATHON INTRO - Sadaf Fardeen, LinkedIn
4:25 – 4:30 – BREAK
4:30 – 5:30 – System Setup Instructions and Team up
5:30 PM - DISPERSE
Wednesday, March 13
8:00am – 4:00pm – HACKATHON (Breakfast included)
12:00 - 1:00 - LUNCH
4:00 - 5:00 - HACKATHON PRESENTATION AND AWARD
Instructions for Hackathon:
Put on your thinking caps, team up with friends, and get ready to be the geek coder!! As a part of SONiC Workshop we are organizing a Hackathon for SONiC. Here are more details of the event. Feel free to reach out with questions if any.
Rules
1. Participate alone or in teams up to a size of 4.
2. Team up with anyone in the community.
3. Free to choose any idea as long as it is useful for SONiC. Some suggested areas: test framework, debugging, performance improvement, standard interfaces, Monitoring.
4. Execute on a virtual switch or a physical switch.
5. Hackathon begins at 4PM on 12th March and ends at 4 PM on 13th March. You are free to work at home or hotel during this period and come back at 4 PM for final presentation.
6. Participants from remote/video conferencing can also join.
7. Final presentation at 4PM on 13th March includes a working demo of 7-10 min.
8. Bring your own laptop for development.
9. Development in any programming language is fine as long as it can integrate with SONiC ecosystem.
10. Use of any off the shelf tools and package is allowed as long as it has appropriate license for use with SONiC.
Instructions
1. Prepare your laptop for virtual switch based development by following instructions given here:
https://github.com/Azure/sonic-mgmt/blob/master/ansible/doc/README.testbed.VsSetup.md
2.Reach out to community and form a team before the event.
Awards
1. Three winning teams to be chosen. $1000 per team.
- Zeus Hacker: Award for most difficult project
- Athena hacker: Award for best idea
- Prometheus hacker: Award for best presentation
2. Judging committee consists of Senior Engineering leaders from SONiC community and OCP
- Haiyong Wang, Senior Director, Alibaba Group
- Lihua Yuan, Partner Developer Manager, Microsoft
- Omar Baldonado, Director of Software Engineering, Facebook
- Shawn Zandi, Director of Engineering, LinkedIn
3. Criteria for evaluation : Relevance for SONiC, complexity in execution, overall presentation and uniqueness of idea.
4. All participants will get a swag

Click here to pre-register. 


Wednesday March 13, 2019 8:00am - 5:00pm PDT
LinkedIn Building R

1:00pm PDT

Early Registration Open
Wednesday March 13, 2019 1:00pm - 6:00pm PDT
The Hub
  Registration
  • YouTube Link -

5:00pm PDT

Data Center Tour - featuring an OCP standard pod designed by Rittal (Offsite) - Must Pre-Register!
What: Join us for an exclusive evening of networking, sharing best practices, and seeing the future of data center design. Key topics include:
  • Best practices in securing your IT Infrastructure
  • High-density in connectivity-rich edge solutions
  • Data center tour, featuring an OCP standard pod designed by Rittal
When: March 13th at 5:00pm - 7:00pm PST.
How: Rittal & Cyxtera invite you to register for the tour, by filling out the form on the right. All attendees will be automatically entered into a drawing to win a bluetooth speaker.
You will receive an email confirmation containing all the details of the event. Buses will run to and from the Convention Center to the Data Center every 30 minutes beginning at 5:00pm and ending at 7:00pm. We hope to meet you at our VIP event and at the show!

Click here to pre-register.


Wednesday March 13, 2019 5:00pm - 7:00pm PDT
Rittal
 
Thursday, March 14
 

7:30am PDT

Breakfast
Thursday March 14, 2019 7:30am - 9:00am PDT
Lobby
  Meals
  • YouTube Link -

7:30am PDT

Registration Open
Thursday March 14, 2019 7:30am - 7:00pm PDT
The Hub
  Registration
  • YouTube Link -

7:30am PDT

OCP Experience Lab
Thursday March 14, 2019 7:30am - 7:30pm PDT
Lobby

7:30am PDT

Future Technologies Symposium - Posters Display
Thursday March 14, 2019 7:30am - 7:30pm PDT
Lobby
  Symposium
  • YouTube Link -

8:00am PDT

Open Source Firmware Meetup
Firmware Developers' Community Meetup

Speakers
avatar for Sai Dasari

Sai Dasari

Software Engineering Manager, Facebook


Thursday March 14, 2019 8:00am - 5:00pm PDT
Marriott Salon V&VI
  EW: OSF/Security
  • YouTube Link -

9:00am PDT

9:14am PDT

Hardware Innovations for Data Growth Challenges at Cloud-scale - Presented by Microsoft
Keeping up with the rapid growth of data, processing it and storing it at reasonable costs are very well documented challenges to all suppliers given the slow growth rates in storage capacities and performance. The major challenge is to engineer solutions that address these issues and work with the OCP community for the benefit of the entire ecosystem. Microsoft Azure will unveil innovations in the storage system to provide high-performance data compression algorithms that reduce costs and optimize overall performance.

Speakers
avatar for Kushagra Vaid

Kushagra Vaid

General Manager and Distinguished Engineer, Microsoft Azure, Microsoft Corporation
Kushagra Vaid is the General Manager for Server Engineering in Microsoft’s Cloud & Enterprise division, and is responsible for driving R&D, engineering and operations for server infrastructure deployed across all of Microsoft’s global cloud scale services. The engineering effort... Read More →


Thursday March 14, 2019 9:14am - 9:34am PDT
Grand Ballroom

9:35am PDT

CEO Welcome Address
Speakers

Thursday March 14, 2019 9:35am - 9:40am PDT
Grand Ballroom

9:41am PDT

OCP Infrastructure for Workload Optimization - Presented by Intel
Since the inception of OCP, the workloads and deployment models driving datacenter infrastructure have changed dramatically. Homogenous infrastructure is being replaced by workload optimized systems. This is driving a shift in datacenter infrastructure to support efficiency through techniques such as resource pooling and open management.  OCP board member and Intel Corporate Vice President, Jason Waxman will discuss how Rack Scale Design and Intel’s latest OCP contributions are enabling datacenter infrastructure to optimize for growing workloads such as AI.

Speakers
avatar for Jason Waxman

Jason Waxman

Corporate Vice President, Intel
Jason P. Waxman is a corporate vice president in the Data Platforms Group. Jason focusses on driving consistent, strategic long-term business planning, accelerating new business models, and identifying inorganic growth opportunities in close collaboration with Intel Capital. Waxman... Read More →


Thursday March 14, 2019 9:41am - 10:01am PDT
Grand Ballroom

10:01am PDT

SONiC/SAI and It's Rapid Growing Ecosystem
The SONiC (Software for Open Networking in the Cloud) community has been growing rapidly since the project was open sourced to OCP in 2016. SONiC provides high agility, availability,  extensibility, and manageability to cloud operators to build hyper scale network. In this keynote, Dave will talk about the evolvement of the project, key contributions and  roadmap of the project. Jointly a few key partners will talk about how challenges in networking, operation can benefit from SONiC and their vision of the ecosystem.

Speakers
avatar for Dave Maltz

Dave Maltz

Distinguished Engineer, Microsoft
Dr. David A. Maltz leads Azure's Physical Network team, which is responsible for developing, deploying, and operating the software and network devices that connect the servers of Microsoft's largest services, including the Azure Public Cloud and Bing.  We write the code for the software... Read More →


Thursday March 14, 2019 10:01am - 10:26am PDT
Grand Ballroom

10:27am PDT

Accelerating Infrastructure – together - Presented by Facebook
The Open Compute Project was started in 2011 to make datacenters more efficient, and help engineers collaborate, share their innovations, and develop common solutions to industry problems, faster. Recently, what's driving our biggest sources of inefficiency and demand has begun to shift. AI, machine learning, video transcode, and advances in network and data fabrics are exposing new problems and opportunities. Coincidentally, many of our systems are changing in fundamental ways. Our industry's engineers are working on these problems, but often in ways that fragment our resources and capacity to innovate, creating new forms of “Gratuitous Differentiation.” With our announcements in 2019, Facebook is seeking to provide platforms for new innovation around which our industry's engineers and other resources can converge. It is our hope that we can continue to grow a mass of developers towards these platforms, so that we can share our innovations, and you can share ours. Let's tackle these new challenges—new platforms, engineering efficiency, and datacenter efficiency—together.

Speakers
avatar for Vijay Rao

Vijay Rao

Director Technology Strategy, Facebook
Vijay Rao is the Director of Technology Strategy at Facebook and is responsible for driving the strategy for areas of infrastructure technology. Facebook builds efficient and flexible infrastructure that scales to accommodate more than a billion people from around the world using... Read More →


Thursday March 14, 2019 10:27am - 10:47am PDT
Grand Ballroom

10:48am PDT

Summit Highlights
Speakers
avatar for Bill Carter

Bill Carter

CTO, OCP
Bill is CTO for the Open Compute Project Foundation where he is a visionary and pathfinder on this open source journey. Bill works with the entire OCP Community and many other software groups, standards organizations, and industry consortiums to bring together people from around the... Read More →


Thursday March 14, 2019 10:48am - 11:00am PDT
Grand Ballroom

11:00am PDT

Expo Hall Open
Thursday March 14, 2019 11:00am - 7:30pm PDT
Expo Hall 1
  Expo Hall
  • YouTube Link -

11:30am PDT

TCO of Innovation in the Data Center - Presented by Western Digital
In the world of scale-out cloud services, whether you are building the next hiking app, bicycle ride-sharing app, or expense-tracking app for corporate travelers, minimizing or reducing your total cost is important. There are multiple ways to achieve data center footprint reduction. New innovations in  technology and continuing advances in 3D NAND are among them. In this talk we'll focus on new innovations in the data center that can help to reduce OpEx and CapEx, all while increasing performance, eliminating critical bottlenecks, and staying up to date with the newest data center innovations.

Speakers
CL

Clint Ludeman

Western Digital


Thursday March 14, 2019 11:30am - 11:45am PDT
Expo Hall 1

11:50am PDT

Future Memory and Storage Technologies Fueling Innovation in the Cloud - Presented by Samsung
Samsung, a key driver of new flash and memory technologies, is the industry leader in NVMe SSDs and high capacity memory. Learn how Samsung’s next generation products will facilitate continuous innovation in the cloud. Topics will include future high density memory solutions, new SSD products that enable the all NVMe data center, as well as intelligent flash storage concepts that will help overcome CPU limitations and maximize data center efficiencies.

Speakers
AG

Anders Graham

SSD Product Marketing Manager, Samsung
AT

Anson Tsui

DRAM Product Marketing Manager, Samsung


Thursday March 14, 2019 11:50am - 12:05pm PDT
Expo Hall 1

12:00pm PDT

Lunch
Thursday March 14, 2019 12:00pm - 2:00pm PDT
Lobby
  Meals
  • YouTube Link -

12:10pm PDT

Open Source Software on White Box - Presented by Delta Electronics
As the industry moves towards an open source software model for networking, candidate technologies such as SONiC and Stratum are emerging as the front runners to deliver a new networking stack for end-customer deployment. Delta is proud to share the status of our work on these technologies to deliver these differentiating technologies to the market.

Speakers
avatar for David Fermor

David Fermor

VP Engineering, Delta
For decades I have worked within great teams to deliver best in class networking products and solutions, working for companies like Google and Cisco. In the last few years I have focused my time on the disaggregated Whitebox networking model, bringing together best in class open HW... Read More →
JF

John Fan

Prinicpal Software Engineer, Information & Communications Technology Business Group (ICTBG), Delta Electronics
TC

Tiffany Chiang

Senior Software Engineer, Information & Communications Technology Business Group (ICTBG), Delta Electronics


Thursday March 14, 2019 12:10pm - 12:25pm PDT
Expo Hall 1

12:30pm PDT

Next-Gen and Breakout Data Center Technologies for the Cloud and Edge - Presented by Wiwynn
The evolution of data centers has been led by different cloud services and the need for enhancing basic infrastructures, actively deploying existing hardware resources, and dynamically configuring workload optimization. Now 5G and Edge computing are entering the breakout phase, where dynamically configurable infrastructures and software-defined data center technologies will play a key role in determining whether or not data centers can meet these emerging demands. From the latest cutting-edge OCP/Project Olympus/Open19 technologies to AI, Edge, 2-phase immersion cooling, and dynamically configurable infrastructure software solutions, come see Wiwynn’s full positioning of next-generation + emerging data center technologies for this breakout year.

Speakers
ES

Ethan SL Yang

Deputy Manager, Wiwynn


Thursday March 14, 2019 12:30pm - 12:45pm PDT
Expo Hall 1

1:00pm PDT

Introduction to Lenovo's Hyper scale Segment - Presented by Lenovo
Presentation by our Lunch Sponsor, Lenovo!

Speakers
avatar for Jeff Sturtz

Jeff Sturtz

Sr. Hyperscale FAE, Lenovo
As a Senior FAE Jeff has the privilege of collaborating with the largest Tier 1 CSPs in the industry by helping them solve their unique business challenges with hardware. Jeff has over 10 years’ experience helping his clients scale their businesses using off the shelf, semi-custom... Read More →


Thursday March 14, 2019 1:00pm - 1:15pm PDT
Expo Hall 1
  Expo Hall Talk
  • YouTube Link -

1:00pm PDT

OCP Impact Study 2019: Market Place Matures, Adoption Increases
At the close of 2018, we conducted a follow on to our 2018 survey of top OCP vendors and end users.
Combined with IHS Markit’s continual industry research, we determine growth of 2018 non-board
member adoption over 2017. In addition, we asked respondents to share with us how the market has
changed over the last 12 months covering technology demands, go to market models, distribution
models, and more. Please join us to learn the details of our findings where we will discus how the
market has matured, review 2018 results vs last year’s forecast and provide an updated forecast to
2022.

Speakers
avatar for Vladimir Galabov

Vladimir Galabov

Principal Analyst, Omdia (formerly IHS Markit)
As a principal analyst in the cloud and data center research practice, Vlad focuses on data center compute covering disruptive trends such as the shift to cloud computing, innovations in the Ethernet adapter and server markets, the adoption of Open Compute Project (OCP) equipment... Read More →
avatar for Cliff Grossner

Cliff Grossner

Executive Director Research & Technology Fellow, Omdia (formerly IHS Markit)
"As head of the Cloud and Data Center Research Practice, he is responsible for research quality across the practice and his research is focused on cloud services, data center compute and networking, and software defined networking (SDN). Dr. Grossner has more than 25 years of telecommunications... Read More →


Thursday March 14, 2019 1:00pm - 1:25pm PDT
210 BF

1:00pm PDT

OCP NIC 3.0: A Collaborative Industry Effort to Create a New Server Connectivity Standard - presented by Intel
In creating the original OCP mezzanine specification in 2012, Facebook sought to define a standard for cost-effective, high-density server connectivity within the OCP framework. In the years since that introduction, the OCP NIC project has gained momentum, with broad industry participation and OCP-NIC-based cards available from most server and Ethernet vendors. While server OEMs continue to support custom cards to enable specific features, recent advancements within the OCP NIC project provide the industry with an opportunity to standardize on a common form factor.

The OCP NIC 3.0 specification, written and in the process of being ratified by the OCP committee members, includes support for advanced network controller features, including additional power, PCIe Bifurcation, and enhanced manageability and generally incorporates key leanings from earlier versions of the specification. With contributions from industry leaders including Facebook, Intel, server OEMs, and Ethernet vendors, OCP NIC 3.0 will simplify adoption of a common mezzanine form factor in server solutions including Cloud, Enterprise, Comms, and others.

Speakers
BB

Brad Burres

Senior Principal Engineer, Chief SmartNIC Architect, DCG Architecture, Intel


Thursday March 14, 2019 1:00pm - 1:25pm PDT
210 DH

1:00pm PDT

PayPal Use Case: Deploying Network Services in a Multi-vendor Enterprise
Core Network Services provides network connectivity for PayPal's site infrastructure. To address 5 9s’ site reliability goals, the demand for the network to perform at its best keeps growing. PayPal transformed its data center networks by deploying white-box leaf/spine designs to help with horizontal scaling, low latency, and resilience to failures. By developing a common framework layer, PayPal is able to address network automation/service API needs in a multi-vendor environment. PayPal will discuss the design approach, lessons learned and how it is leveraging various vendor technologies in its transformation journey.

Speakers
SK

Saikrishna Kotha

Director of Network Engineering, PayPal


Thursday March 14, 2019 1:00pm - 1:25pm PDT
210 CG

1:10pm PDT

Inspur RMC Technology Simplifies the Management in OCP Datacenters - Presented by Inspur
Management has always been a challenge in data centers. Based on the Open BMC technology that Inspur introduced in last year, we will apply that technology into the Rack level management utilities. Alan will elaborate Inspur solution and key benefits to broad audiences and OCP adopters.

Speakers
avatar for Alan Chang

Alan Chang

Sr. Product Director, BU Cloud Computing, Deputy GM, IEI Server Product Line Dept., Inspur
With over 10+ years’ experience in various roles in server product planning & account management in OxM. Having worked as everything from software programmer fresh out of university to hardware product planning in major system supplier, Alan has developed a diverse skillset that... Read More →



Thursday March 14, 2019 1:10pm - 1:25pm PDT
Expo Hall 1

1:30pm PDT

Experience Data Together - Presented by Seagate
As demand for data continues to explode, the global Datasphere will balloon to 175 zettabytes by 2025. In this session, Seagate will discuss the central role the open compute community must play to promote breakthroughs in recording physics and innovations in areal density technology to support this growth. At the same time, the speed of access to data must continue to scale to improve interaction between the storage and compute tiers. And as new applications and devices at the edge further ramp up data creation, the data infrastructure must move data securely and with velocity. It’s vital for the storage OCP community to take an active role to define pain points for the storage ecosystem, and provide solutions that optimize data's value.

Speakers
JF

Jason Feist

CTO Office, Seagate


Thursday March 14, 2019 1:30pm - 1:55pm PDT
210 DH

1:30pm PDT

Intel Cloud Innovations, An Overview - Presented by Intel
This session will provide an overview of the Cloud related innovations from Intel in the areas of Si, Platforms, Firmware, Connectivity. This session will also highlight partnership work with other cloud partners towards OCP. Network portfolio and product details will also be discussed.

Speakers
avatar for Mohan J Kumar

Mohan J Kumar

Fellow, Intel
Mohan Kumar is an Intel Fellow in the Data Platforms Group and the director of Cloud Architecture and Solutions in the Cloud & Enterprise Solutions Group at Intel Corporation. His organization is responsible for driving the architecture requirements for Intel Cloud customers.Kumar... Read More →


Thursday March 14, 2019 1:30pm - 1:55pm PDT
210 CG

1:30pm PDT

Scaling to a Larger Bandwidth Network with MiniPack - Presented by Facebook
This session will go over hardware, software, network architecture, and operational characteristics associated with a new Facebook network fabric design, and Facebook's new fabric switch: Minipack. It include insights on implementation and migration (green and brownfield), and updates on how Facebook our industry's future challenges scaling to larger, higher bandwidth networks, and with network heterogeneity. MiniPack is Facebook's 2nd generation fabric switch, designed for new topologies and emerging challenges in high bandwidth networking.  It features lower power, fewer switching elements, and a modular arcitecture that supports multiple port interface types, as well as design in-mind for more tightly integrated optical interfaces, in the future.

Speakers
avatar for Omar Baldonado

Omar Baldonado

Director of Software Engineering, Infra Network, Facebook
Omar Baldonado is the co-lead of the OCP networking project. At Facebook, he is part of the networking team, responsible for large-scale software systems that model, automate, monitor, and control Facebook’s network. The network domains include Facebook’s data centers; global... Read More →


Thursday March 14, 2019 1:30pm - 1:55pm PDT
210 BF

1:50pm PDT

Xilinx Alveo Adaptable Acceleration Solutions for the Data Center - Presented by Xilinx
Heterogeneous computing in the data center is here to stay. What the hyperscale cloud service providers have popularized with FPGA-based solutions is now available to everyone. Xilinx is revolutionizing the data center by providing adaptable, high performance platforms for workload acceleration. See what Xilinx has up its sleeve with its new Alveo acceleration cards as it reveals an expanded set of application partners and use cases. Then come see us at our booth for live demonstrations.

Speakers
JD

Jim Dworkin

Director of Data Center Marketing, Xilinx


Thursday March 14, 2019 1:50pm - 2:05pm PDT
Expo Hall 1

2:00pm PDT

OCP Security Project Meetup
Look for "OCP Security Project" Table Top signs on round tables in the lobby area of the LOWER LEVEL.

Speakers
avatar for Nate Klein

Nate Klein

OCP Security Project Lead, Google
Nate Klein is a hardware engineer on the Platforms Infrastructure team that designs and deploys all of the hardware that powers Google's internal and cloud services.  In 6 years at Google his work includes NIC drivers and firmware, server and PCIe card design, and hardware security... Read More →



Thursday March 14, 2019 2:00pm - 3:00pm PDT
Lower Level - Tables in Open Area - See Map

2:05pm PDT

New Infrastructure Architectures for Optimizing the Modern Data Center - Presented by Marvell
As connected intelligence evolves at an astonishing rate, the data centers processing, moving, storing and securing all this data are also morphing in many ways. Data centers can no longer take a “one size fits all” approach to their architecture and technologies. As compute moves to the edge, data centers must address composability and workload-specific requirements, among others. This presentation will look at the innovative architectures to bring efficiencies and scale to meet these requirements and future-proof the infrastructure. The session will discuss how advanced storage, networking and compute silicon building blocks are optimizing the modern data center, highlighting specific use cases that evaluate considerations such as capacity, latency, performance, power and cost, and the value propositions for the OCP community.

Speakers
NA

Nigel Alvares

VP of SSD and Data Center Storage Solutions, Marvell


Thursday March 14, 2019 2:05pm - 2:30pm PDT
210 DH

2:05pm PDT

OCP Design for Huawei Cloud Infrastructure - Presented by Huawei
Huawei is rapidly expanding our global public cloud service and on-premise cloud and IT infrastructure. Inspired by the great contributions by various OCP participants in past few years, Huawei has stepped up to develop an optimized rack scale design that will become the foundation of our cloud and IT infrastructure roll out. This presentation shares the details of our approach and design considerations


Speakers
avatar for Jinshui Liu

Jinshui Liu

CTO, IT Hardware Domain, Huawei Technologies
Chief Architect for IT Hardware at Huawei & the Director of D-Lab, currently overseeing the development of easy to use AI training and Inferencing systems based on Nvidia GPU and Huawei in-House Developed AI-ASIC;The Chief Architect of Huawei Atlas Heterogeneous Computing Platform... Read More →


Thursday March 14, 2019 2:05pm - 2:30pm PDT
210 BF
  Executive Track
  • YouTube Link -

2:05pm PDT

Software-defined Flash Futures Driving Next-Generation Cloud Services- Presented by Microsoft
As the line blurs between traditional hardware and software roles in the cloud, A new class of flexible Flash/SSD storage architectures has emerged delivering intelligence, application acceleration and value services like compression and security. This talk outlines the benefits of this exciting new  type of storage to cloud services and hyper-scale applications. Areas include Project Denali, Computation SSDs and other compute-in-memory technologies and how they are changing cloud computing over the next decade.

Speakers
avatar for Michael Cornwell

Michael Cornwell

General Manager, Storage Technologies, Azure Hardware Infrastructure, Microsoft


Thursday March 14, 2019 2:05pm - 2:30pm PDT
210 CG

2:30pm PDT

Open SeaChest, a Highly Portable Drive Management Project - Presented by Seagate
Seagate will discuss an open source cross-platform project containing a collection of utilities written for storage products based on standard SAS, SATA, and NVMe specifications, allowing device diagnostics and management. The solution is portable, secure and non-disruptive, and integrates in diverse hardware platforms & operating systems to enable storage device management.

Speakers
MA

Muhammad Ahmad

Sr. Engineering Manager, Seagate


Thursday March 14, 2019 2:30pm - 2:45pm PDT
Expo Hall 1

2:40pm PDT

Common Building Blocks Product Architecture and Technology for Workload Optimization: Presented by Wiwynn
The Internet/Cloud applications are getting diversified, while each Cloud Service provider embraces its own approaches in building its Data Center infrastructure. Therefore, rare products fit majority of CSPs requirements in optimizing the workload performance while keeping the Total Cost of Ownership (TCO) low. To achieve the goal of optimizing the workload performance while keeping the TCO low. Wiwynn’s strategies are: . Design products with Common Building Blocks approaches among different Rack Infrastructures to minimize investment and expedite time to deployment. . Invest in effective thermal solution, high power conversion efficiency, and high speed signal design technologies to enable workload optimized product design and to reduce the TCO.

Speakers
SL

Steven Lu

Vice President of Product Management and Marketing, Wiwynn


Thursday March 14, 2019 2:40pm - 3:05pm PDT
210 BF

2:40pm PDT

SONiC and Self-healing Network - Presented by LinkedIn
LinkedIn is using SONiC as the network operating system to implement OpenFabric control plane and the vision of self-healing networks. In addition, LinkedIn is working with the SONiC community to define standardized API, to run some of the same infrastructure software and tools used on application servers including Kafka, telemetry, and software development toolkits.

Speakers
avatar for Shawn Zandi

Shawn Zandi

Head of Network Engineering, LinkedIn
Director of Engineering, leading all network engineering at LinkedIn. I oversee new hardware and software developments including network architecture, deployments and operations, from global production data centers to backbone and edge.


Thursday March 14, 2019 2:40pm - 3:05pm PDT
210 CG

2:40pm PDT

Xilinx Advances in Data Center Acceleration Solutions - Presented by Xilinx
It is an exciting and busy time to be offering acceleration solutions for data center use cases. Not only is the explosion of data overwhelming the compute capacity of traditional CPU-based clusters, but also the lesser known slowing of Koomey’s Law leads to decreased power efficiency over time. Xilinx and its partners are revolutionizing the data center by providing adaptable, high performance and power efficient solutions for compute, storage and network acceleration. In this presentation, we will highlight available and deployable solutions for the data center, both in the cloud and on premises, enabled by the Xilinx Alveo portfolio of acceleration cards as well as our growing ecosystem of hardware and software partners.

Speakers
RS

Ravi Sunkavalli

VP of Data Center Marketing, Xilinx
MM

Manish Muthal

VP of Data Center Marketing, Xilinx


Thursday March 14, 2019 2:40pm - 3:05pm PDT
210 DH

2:50pm PDT

High Performance 64-bit eMAG Arm Server Processor For Cloud and Edge Computing - Presented by Ampere Computing
Cloud and edge computing workloads typically require a mix of high compute performance, memory bandwidth and capacity, and I/O throughput, as well as a strong ecosystem of software and hardware vendors. Ampere’s recently launched 64-bit Arm eMAG™ server processor delivers a balance of compute, memory, and I/O performance in a single integrated device, enabling it to excel in a range of applications, including Web tier, databases, storage, and big data.  This presentation highlights eMAG’s value proposition for these applications and provides an update on Ampere and its expanding partner ecosystem.

Speakers
avatar for Surya Hotha

Surya Hotha

Vice President, Product Management and Marketing, Ampere Computing
Surya Hotha is the vice president of product management and marketing at Ampere. He brings 20 years of experience in the semiconductor industry with primary focus on networking and server infrastructure. Prior to joining Ampere, Surya served as the head of product management and marketing... Read More →


Thursday March 14, 2019 2:50pm - 3:05pm PDT
Expo Hall 1

3:10pm PDT

Concept Introduction for OCP Next Generation Rack and Chassis – Tai Chi - Presented by QCT
As the key components in servers are all trending toward higher power consumption, QCT thinks it is the right moment to revisit basic system architecture for a next generation OCP chassis. Moreover, how to solve rack level manageability is another key subject that we would like to tackle together with this new design. By leveraging existing OCP standards, QCT would like to introduce how we envision the concept of an OCP future system and rack that brings a convergence of Open Rack and Project Olympus.

Speakers
JL

Jack Luoh

Product Manager, QCT


Thursday March 14, 2019 3:10pm - 3:25pm PDT
Expo Hall 1

3:15pm PDT

Hewlett Packard Enterprise, A Fresh Approach to Open - Presented by HPE
From Hewlett Packard Enterprise, the company that created industry-standard compute comes a fresh approach to Open Compute. Explore the benefits of a global supply chain, a comprehensive product portfolio and industry-leading consulting services designed for aspiring enterprises of all sizes.

Speakers
TL

Tom Lattin

Vice President, ProLiant and Cloudline Systems Product Management, HPE


Thursday March 14, 2019 3:15pm - 3:40pm PDT
210 BF

3:15pm PDT

Project Olympus 3U PCIe Expansion Server Deep Dive - Presented by ZT Systems
The Project Olympus building blocks provide an excellent foundation to develop and configure server solutions for specific server workloads (compute, cold storage, performance storage, and AI). This session will provide the technical details on the 3U PCIe expansion system ZT developed (including thermals, PDB, and riser cards) using the Project Olympus Building Blocks. In addition, performance data will be presented on several GPU configurations.

Speakers
RM

Raymond Miles

VP of Architecture, ZT Systems


Thursday March 14, 2019 3:15pm - 3:40pm PDT
210 DH

3:15pm PDT

SmartNIC Makes Composable Cloud Storage a SNAP - Presented by Mellanox
The next generation of cloud native applications are based on a container based microservice architecture with dynamic scaling. These applications support a wide variety of workloads, each have differing needs for compute, storage, and networking. Thus, the ideal architecture is highly composable and able to transcend the specific configuration of physical server and storage infrastructures and instead deliver fully software defined, virtualized resources that provides as much compute, storage and networking as needed. Moreover, this flexibility must be provided in a way that is transparent to applications and provides agility, scalability, and secure isolation. This presentation will introduce a new SmartNIC technology that blurs the lines between local and remote NVMe Storage and makes all cloud resources be appear local. The presentation will provide a deep-dive into those new requirements and demonstrate a solution that brings virtualized storage to bare-metal clouds and delivers disaggregation of compute and storage to achieve fully optimized resource utilization.

Speakers
avatar for Kevin Deierling

Kevin Deierling

VP Marketing, Mellanox Technologies
Kevin Deierling is VP of marketing at Mellanox Technologies, responsible for enterprise, cloud, storage, big data, and machine learning businesses. Previously he served as VP of technology at Genia Technologies, Chief Architect at Silver Spring Networks, and VP marketing and business... Read More →


Thursday March 14, 2019 3:15pm - 3:40pm PDT
210 CG

3:30pm PDT

Unlocking the Power of Circular Datacenters - Presented by IT Renew
Billions of dollars are left on the table every year in unrealized TCO savings, and billions of people and opportunity remain off line globally.   While open source software and open hardware are helping to democratize access and efficiency… we have a long way to go.   Circular data centers are the answer - maximizing the lifetime value of IT assets, the global scaling of downstream use cases, and - conveniently - also sustainability.  Orchestration of circular models requires a new way of thinking - from design to decommissioning and redeployment, new business models and products, and global collaboration around open standards and a shared ethos.  In this session we reveal both the operational blueprint and the new open hardware product line that together stand to activate global demand for open hardware.  Whether you are one of the few, focused on absolute returns at massive scale, or one of the many, scrapping to grow in the face of constraints, we will help you with your imperative: minimize TCO, achieve competitive advantage, and be a part of the global sustainability solution.  

Speakers
avatar for Ali Fenn

Ali Fenn

President, ITRenew
Ali Fenn is the President at ITRenew, where she oversees all revenue and leads the Company’s circular data center initiatives, including market development and business model innovation. Open hardware platforms and open source software innovation are the critical foundation underpinning... Read More →


Thursday March 14, 2019 3:30pm - 3:45pm PDT
Expo Hall 1

3:50pm PDT

Facebook AI Infrastructure- Presented by Facebook
AI/ML is essential to many core products and services at Facebook. This talk covers the hardware and software infrastructure that supports Facebook AI at global scale. Facebook’s AI workloads are extremely diverse with services requiring many different types of models in practice. This diversity has implications at all layers of the system stack. In addition, computational requirements are also intense, leveraging both GPU and CPU platforms for training and abundant CPU capacity for real-time inference. Addressing these and other emerging challenges continues to require diverse efforts while considering hardware and software co-design.

Speakers
avatar for Whitney Zhao

Whitney Zhao

Engineer, Facebook
Whitney is a seasoned hardware engineer leading AI/ML system design in Facebook. Whitney has led multiple hardware generations ranging from general purpose 2S system such as Tioga Pass to ML JBOG Big Basin systems, all of which have been contributed to OCP. She has been driving multiple... Read More →
SN

Sam Naghshineh

Technical Program Manager, Facebook


Thursday March 14, 2019 3:50pm - 4:05pm PDT
Expo Hall 1

3:50pm PDT

Application of AI Scale-Out Inferencing - Presented by QCT
As AI becomes more pervasive in every facet of our life, we need to start facilitating the transition to these new types of workloads. In this session, James Jau, VP of Research and Development at QCT, will discuss Scale-Out Inference Architecture applications and what can be achieved as we move towards an accelerated future.

Speakers
JJ

James Jau

VP of Research and Development, QCT


Thursday March 14, 2019 3:50pm - 4:15pm PDT
210 DH

3:50pm PDT

Open Computing Reduces the TCO and Accelerates Adoption of OCP, esp. for Small and Medium Adopters - Presented by Inspur
In addition to hyper-scale DC customers, CIO of small and medium customers has been suffering from how to reduce TCO while not being manipulate by large OEM vendors. Inspur looks on OCP as a good vehicle. Wilson will introduce how Inspur continuously meet the different types of users’ needs from hardware system, and various application scenarios, to open management technology.  

Speakers
avatar for Lillian Wang

Lillian Wang

SVP/GM of Internet Industry BU, Inspur Group, Inspur
Lillian Wang is the Senior Vice President of Inspur Group and General Manager of the Internet Industry BU.  She is responsible for strategic planning, product strategy and product development for internet , energy and transportation segments.  Lillian joined Inspur in 1997 and has... Read More →
avatar for Wilson Guo

Wilson Guo

Technology Director, Inspur


Thursday March 14, 2019 3:50pm - 4:15pm PDT
210 BF

3:50pm PDT

Solving the Data Center Energy Crisis: Creating Energy-Efficient, High-Performance Servers - Presented by Ampere Computing
Data centers consume more than 2 percent of the world’s electricity. Global power consumption is expected to triple in the next five years.
While these statistics are alarming, the demands for hyperscale platforms requiring performant workloads and lower power cost savings will continue to rise. 
These requirements are dependent on the shift within the industry since the start of OCP to require open infrastructure and open standard software to drive
these platforms, while also delivering energy efficiency. 
 
Mauri Whalen, VP of Ampere Computing will discuss how Ampere is aligning with OCP infrastructure standards to deliver next generation platforms
that are high-performance, low power and energy efficient.

Speakers
MW

Mauri Whalen

Vice President Software Engineering, Ampere Computing


Thursday March 14, 2019 3:50pm - 4:15pm PDT
210 CG

4:10pm PDT

The State of Hardware Security: Cerberus Present and Future - Presented by Microsoft
Microsoft announced Project Cerberus in October 2017 at the Open Compute Project Summit and contributed the specifications to OCP community.  In this presentation, we will discuss Cerberus V1 development and products and go over Cerberus Next, a proposal for second generation Cerberus for hardware security.  Cerberus Next will provide in-silicon hardware security which will be an enhancement over V1.

Speakers
avatar for Bryan Kelly

Bryan Kelly

Principal Firmware Engineering Manager, Microsoft
Bryan Kelly is a Principle Firmware Engineering Manager for Microsoft’s Azure Cloud Server Infrastructure team. Bryan supports Microsoft’s cloud services by designing and developing firmware that enables hardware solutions in Microsoft’s next generation cloud platforms. During... Read More →
avatar for Badriddine Khessib

Badriddine Khessib

Sr. Director, Microsoft


Thursday March 14, 2019 4:10pm - 4:25pm PDT
Expo Hall 1

4:25pm PDT

Memory Storage Innovations Fueling the Public Cloud - Presented by Samsung
Cloud datacenters in 2019 should look forward, not backward. Restrictive legacy technologies developed for spinning hard disks prevent us from taking full advantage of the advanced performance and physical capabilities of solid-state storage. Movement of data from processing has traditionally caused increased latency and energy consumption. Intelligent storage such as Samsung’s SmartSSD will overcome these issues by incorporating an FPGA accelerator into the SSD unit, resulting in higher processing performance and improved time-to-insight. NVMe SSDs provide big speed boosts and new architectural flexibility compared to the aging SATA standard, while PCI-Express Gen 4 offers the “bigger pipe” and scalable innovation needed for massive data loads. And system architects and operations teams alike will appreciate flash native form factor SSDs that use space efficiently and enable key enterprise features like hot-swap, power loss protection, and front drive loading.  This session will take a closer look at memory storage innovations that are fueling today’s public cloud.

Speakers
avatar for Michael Smullen

Michael Smullen

Sr. Manager Datacenter SSD Products Group, Samsung
Cradle to grave product owner: definition, MRD/PRD, development, rollout, early customer programs, pricing strategy and tactical negotiation, design win budget achievement, revenue and gross margin management, and EOL analysis and execution.Specialize in technology products in the... Read More →


Thursday March 14, 2019 4:25pm - 4:50pm PDT
210 DH

4:25pm PDT

New Era of Data Center Networking Technology – Intelligence-defined Optical Tunnel Network System (OPTUNS) - Presented by Delta Electronics
Delta Electronics has over ten years of experience in data center solutions. A promising datacenter networking solution is deemed to be one of the cornerstones to making 5G promises a reality. In this talk, Maria Yuang first addresses several main challenges of building a viable datacenter network infrastructure that cannot be achieved with the typical technologies in some use cases. She will review the current trend in SDN-based paradigm in the context of datacenter network control.
Dr. Yuang will then present the pioneering research and development of an intelligence-defined optical data center network infrastructure by Delta and NCTU, called Optical Tunnel Network System (OPTUNS). OPTUNS consists of two types of optical switching nodes, and is controlled based on an SDN+ paradigm. The talk will showcase the OPTUNS architecture, the prototypes of optical switching nodes, and the proof-of-technology testbed. The talk will also highlight some key performance benchmarking results.

Speakers
MC

Maria C. Yuang, Ph.D

Director, Delta-NCTU R&D research center, Delta Electronics


Thursday March 14, 2019 4:25pm - 4:50pm PDT
210 CG

4:25pm PDT

Operationalizing the Circular Opportunity - Introducing Sesame: Open Hardware Compute and Storage for Everyone - Presented by IT Renew
This session makes the amorphous idea of a circular data center real.  We will demonstrate the potential of circular models in driving asset lifetime value and TCO for major data center operators, and we will showcase the downstream power the model yields in catalyzing access, opportunity, and market expansion for open hardware solutions.   Delivering on this vision requires engineering, quality, logistics, and market-making expertise.  Born from this environment, Sesame is a first-of-its-kind suite of fully recertified and warrantied open hardware products.   The Sesame opportunity is to fuel secondary markets beyond the hyperscale data center arena, and we will showcase customer experiences from a major Telco, the Aerospace industry, and a modern edge computing paradigm that stands to truly democratize access to and efficiency of IT resources.    

Speakers
avatar for Ali Fenn

Ali Fenn

President, ITRenew
Ali Fenn is the President at ITRenew, where she oversees all revenue and leads the Company’s circular data center initiatives, including market development and business model innovation. Open hardware platforms and open source software innovation are the critical foundation underpinning... Read More →
avatar for Aidin Aghamiri

Aidin Aghamiri

CEO, ITRenew
An integral part of ITRenew since 2007, Aidin has built the groundbreaking products and services that have made the company the leading data center decommissioning provider in the world. As CEO, Aidin brings leadership, strategy and technology implementation experience to his rol... Read More →


Thursday March 14, 2019 4:25pm - 4:50pm PDT
210 BF
  Executive Track
  • YouTube Link -http://www.youtube.com/watch?v=FFK9dl-m38I

4:30pm PDT

Cloudline Servers Autonomous Driving Solutions - Presented by HPE
Learn how HPE is accelerating insights through a new generation of data and analytics using open design principles

Speakers
EH

Ed Holden

HPE ProLiant & Cloudline Planning Manager, HPE


Thursday March 14, 2019 4:30pm - 4:45pm PDT
Expo Hall 1

4:50pm PDT

High-Performance Switching Without Compromise - Presented by Marvell
The demand for high bandwidth is reaching new heights, fueled by emerging 5G networks, video surveillance, video streaming, machine learning and artificial intelligence. This demand is causing a massive infrastructure upgrade across hyperscale and private data centers, campus and enterprise networks, and mobile and fixed infrastructure. It is also driving new architectures such as composable or kinetic infrastructure, edge data centers and disaggregated storage. In this session, we will discuss the high-performance ethernet switching requirements for the new architectures in next-generation enterprise and edge data centers. We will share how these requirements are driving advanced features to help bring reduced power, latency and cost to data center infrastructure, and how the implementation of these new ethernet switching devices can bring a new level of optimization across different use cases.

Speakers
avatar for Yaniv Kopelman

Yaniv Kopelman

Networking CTO, Marvell
Yaniv Kopelman is the Networking CTO of the Networking Business Unit (NBU) at Marvell Semiconductor, where he re-joined  the company in June of 2015 after a 12-year tenure from 1996-2008. In his position, Yaniv is responsible for defining the company’s cloud network infrastructure... Read More →


Thursday March 14, 2019 4:50pm - 5:05pm PDT
Expo Hall 1

5:00pm PDT

Accelerating the Evolution of Data Storage - Presented by Western Digital
Cloud applications today are driving evolutionary and revolutionary innovations in hard drives and solid state storage. Storage devices are being tuned to better align with cloud demands of scale, adaptability, and performance. Hard drives are evolving with technologies that increase capacity, provide online re-configurability, and enhance access density. Solid state drives are utilizing the latest in 3D NAND and cloud-friendly form factors to deliver the higher quality of service demanded by the Cloud. Working with our partners to combine these innovations in the data center, Western Digital is enabling the next generation of environments for data to thrive.

Speakers
EB

Eyal Bek

Vice President, Product Marketing, Western Digital


Thursday March 14, 2019 5:00pm - 5:25pm PDT
210 BF

5:00pm PDT

Breaking Barriers in AI: New Hardware, Standard Platforms - Presented by Intel
Industries and companies thrive when the focus is on innovation and expanding capabilities.  Technology standards help foster innovation by providing a foundation to generate seamless experiences for developer and user alike.  Intel is a longtime believer in democratizing technology and making it accessible to all. Our history of contributing to open source projects extends back decades, and we’re taking that same approach with our AI work today.  In this talk, we will discuss how Intel is supporting OCP and industry standards, and how new AI hardware like our Intel® Nervana™ Neural Network Processors, paired with open source software, will allow developers to harness the power of AI for any application.

Speakers
avatar for Julie Choi

Julie Choi

VP and GM of Artificial Intelligence Products and Research Marketing, Intel Corporation
Julie Shin Choi is VP and GM of Artificial Intelligence Products and Research Marketing at Intel Corporation. She is responsible for marketing the Intel portfolio of hardware and software products for building enterprise scale AI solutions. She is driving the AI marketing strategy... Read More →


Thursday March 14, 2019 5:00pm - 5:25pm PDT
210 CG

5:00pm PDT

Deep Learning Infrastructure Challenges: Panel Discussion - Presented by Hyve
With deep learning now a mainstream technology in hyperscale data centers but still developing quickly In various directions, the panel discussion will cover various challenges of deploying infrastructure for that at scale.  By selecting a panel with diverse experience and expertise in this area, we want to cover thought leaders inputs on seamlessly adopting new deep learning hardware, moving data in and out of AI clusters, performance testing and comparisons, as well as practical issues around physical infrastructure and other planning.

Speakers
avatar for Eric Shobe

Eric Shobe

Hardware Engineering Manager, Dropbox
avatar for Andres Rodriguez

Andres Rodriguez

Sr. Principal Engineer, Intel
Andres Rodriguez is a Sr. Principal Engineer in the Data Center Group (DCG) where he designs deep learning solutions for Intel’s customers and provides technical leadership across Intel for deep learning products. He has 15 years of experience working in artificial intelligence... Read More →
avatar for Manoj Wadekar

Manoj Wadekar

Storage Hardware Architect, Facebook
Current Focus: Solve ambiguous problems for hyper-scale infrastructure. Find the problems, drive innovative solutions.Background:28+ years in Server, Storage, and Networking industry.Specialties: Transforming hyper-scale infrastructure to drive performance and efficiency (proven for... Read More →
avatar for Jay Shenoy

Jay Shenoy

VP of Technology, Hyve Solutions
I drive the road map and product planning for Hyve Solutions. Talk to me about any of your digital infrastructure needs!
BM

Balamurali Meduri

Head of Platform AI, eBay
avatar for Matthew Singer

Matthew Singer

Sr. Staff Hardware Engineer, Twitter
Matt Singer is Senior Staff Hardware Engineer at Twitter. An expert in performance and server architecture, Matt is the technical leader for the hardware team that created one of the most resilient and cost-effective cloud infrastructures in the industry. He is also the architect... Read More →


Thursday March 14, 2019 5:00pm - 5:25pm PDT
210 DH

5:10pm PDT

SONiC at 200/400G - Ready for Hyper Scale - Presented by Mellanox
200 and 400 Gigabit Ethernet solutions are now shipping from an ecosystem of optics and switch vendors - but only the most open of these switches support SONiC today. We will review the use cases for these high speed Open Ethernet switches. Besides raw speed, these new 200/400G switches will bring new telemetry and dynamical routing functionality to SONiC powered data centers. Lastly, we will identify some gotchas to look out for as not all 200/400G switches are created equal.

Speakers
avatar for David Iles

David Iles

Sr. Director, Ethernet Switch, Mellanox
David Iles, Senior Director of Ethernet Switching at Mellanox Technologies:Over the last 20 years, David has held leadership roles at 3COM, Cisco Systems, Nortel Networks, and IBM where he pioneered advanced network technologies including High Speed Ethernet, Layer 4-7 switching... Read More →


Thursday March 14, 2019 5:10pm - 5:25pm PDT
Expo Hall 1

5:30pm PDT

Opening Night Happy Hour
Thursday March 14, 2019 5:30pm - 7:30pm PDT
Expo Hall 1
  Meals
  • YouTube Link -
 
Friday, March 15
 

7:30am PDT

Breakfast
Friday March 15, 2019 7:30am - 8:30am PDT
Lobby
  Meals
  • YouTube Link -

7:30am PDT

Registration Open
Friday March 15, 2019 7:30am - 3:00pm PDT
The Hub
  Registration
  • YouTube Link -

7:30am PDT

OCP Experience Lab
Friday March 15, 2019 7:30am - 5:00pm PDT
Lobby

7:30am PDT

Future Technologies Symposium - Posters Display
Friday March 15, 2019 7:30am - 5:00pm PDT
Lobby

8:00am PDT

Project Olympus and Open Rack Converges
Project Olympus systems are based on modular and independent solutions tied into a cloud-scale management utilizing 19" EIA Racks. Open Rack systems have a large following in OCP, but do not yet have the flexibility that Microsoft needs in it's systems. Today we'll present efforts by Microsoft and other OCP companies to converge the two concepts into a new platform that bridges to create the best of two worlds.

Speakers
avatar for Shaun Harris

Shaun Harris

Director Power Systems and Compliance, Microsoft
avatar for Brandon Rubenstein

Brandon Rubenstein

Principal Platform Engineer, Microsoft
I've had the privilege to contribute to Microsoft's success as an engineering director, a design architect and a lead designer. Most recently I've managed the mechanical and power teams responsible for defining, designing, and maintaining Microsoft's cloud business hardware and datacenter... Read More →


Friday March 15, 2019 8:00am - 8:18am PDT
210 BF

8:00am PDT

Multi-actuator HDD Deployment in Hyper-Scale Data Centers
As cloud datacenters have deployed increasingly higher-capacity HDDs, some workloads have saturated the IOPs/TB limitations for these high capacity devices. Seagate has announced a dual actuator HDD (“Mach.2”) that effectively doubles the IOPs/TB for this device. Seagate will overview the architecture, performance, and power characteristics of Mach.2 and applicability in hyperscale environments.

Speakers
JB

James Borden

Principal Cloud Architect, Seagate Technology
TW

Tim Walker

Technologist, Seagate Technology


Friday March 15, 2019 8:00am - 8:18am PDT
212 AB

8:00am PDT

Direct Contact Liquid Cooling Performance on a Project Olympus Server
Direct Contact Liquid Cooling uses the exceptional thermal conductivity of liquid to provide dense cooling to targeted areas. By integrating DCLC, dependence on fans and expensive air handling systems is drastically reduced. This enables higher rack density, overall reduced power use and peak performance potential. In this engineering workshop, CoolIT Systems will quantify the benefits of DCLC by presenting test data from a Project Olympus server. An analysis of the heat captured in liquid, maximum case temperature, and power consumption is presented at different liquid temperature (ASHRAE W1, W2, W3, W4).

Speakers
avatar for Cam Turner

Cam Turner

Director, Hyperscale Solutions, CooIIT Systems
CoolIT Systems Inc. is a world leader in direct contact liquid cooling for the Data Center market. CoolIT's Rack DCC platform is a modular, rack-based, advanced cooling solution that allows for dramatic increases in rack densities, component performance, and efficiencies. This vendor... Read More →


Friday March 15, 2019 8:00am - 8:25am PDT
210 E

8:00am PDT

Data Center Facility Next Gen
The Data Center Facilities project has been successfully reincubated and has gained momentum in its development and launch of the OCP Ready Program and the Modular Data Center Subproject. Moving the Project forward to maximize our contribution to the community requires a defined structure and organization around the project's focus areas. A revised Charter with the new structure and vision for the project will be presented.

Speakers
avatar for Brevan Reyher

Brevan Reyher

Design & Build Engineer/Data Center Facility Project Lead, Rackspace/OCP Foundation
Brevan Reyher leads the Global Data Center Infrastructure Design & Build DC Engineering practice area at Rackspace Hosting. After graduating from Trinity University in San Antonio, he has spent over 10 years in the design, construction, operation and maintenance of data center facilities... Read More →


Friday March 15, 2019 8:00am - 8:25am PDT
211 AB

8:00am PDT

OCP Profiles for Hardware Management
The Hardware Management Project creates designs and specifications to enable inter-operable manageability for Open Compute platforms. This project provides a common foundation for providing manageability for other Open Compute projects. In this talk, we will describe OCP profiles for baseline manageability capabilities across the following Open Compute projects: Compute, Storage, Networking, Telco, Data center facility, HPC, Rack, and Power.

Speakers
avatar for John Leung

John Leung

Principal Engineer, Intel
John Leung is a Principal Engineer at Intel Corporation in the Data Platforms Group. John has been in the computer industry for over 36 years: architecting, designing and coding. At Intel, John is currently focused on the manageability interfaces for data center compute platforms... Read More →
avatar for Hemal Shah

Hemal Shah

Distinguished Engineer, Broadcom
Hemal Shah is a Distinguished Engineer and Systems/Software/Standards architect in the Compute and Connectivity (CCX) division at Broadcom Inc. He leads a team of architects. In his role, he is responsible for the product architecture definition of Ethernet NIC products, NIC integration... Read More →


Friday March 15, 2019 8:00am - 8:25am PDT
Marriott Salon I&II

8:00am PDT

Exceeding the Limits of Current Network Technology by Leveraging the Full Potential of Barefoot Tofino and the P4 Programming Language with the New STORDIS Switch
Barefoot Tofino and the P4 programming language are changing the way network fabrics are built. Current switch hardware based on Tofino have hit the upper bound in regards to running intensive pipelines and are missing certain important components. Hence STORDIS has launched spine and core versatile network switches offering a unique combination of powerful, high quality components and a chip set programmable with P4.
The switches deliver an advanced high-level performance providing a multitude of possibilities, such as in-band network telemetry and intelligent network packet brokers. It is the first Barefoot Tofino chipset supporting the PTP time sync feature and the only Tofino based switch providing 1Gb support to connect existing legacy equipment with future hardware.
STORDIS is a proud OCP Platinum Member and looking forward to share relevant knowledge with the community, to inspire and to support everyone to benefit from the achievements of the Open Networking Revolution.
Working in the IT industry for more than 10 years, STORDIS experienced the evolution of open networking and learned how to handle the increasing demand for capable, flexible and efficient information technology. Harsha Vardhan, Network Systems Engineer of STORDIS, will provide a more in-depth insight into the architecture of STORDIS Advance Programmable Switches and their efficient deployment in productive network environments.

Speakers
avatar for Harsha Vardhan

Harsha Vardhan

Network Systems Engineer of STORDIS,, Stordis


Friday March 15, 2019 8:00am - 8:25am PDT
211 CD

8:00am PDT

Leveraging Open Compute Technologies to Build a Portable NOS
Increasingly, service providers and network operators outside of the data center wish to leverage the advantages of disaggregation that their peers in the data center have enjoyed for several years. This puts pressure on OEMs to supply a portable NOS that runs on Open Compute compatible hardware as well as their own. We will examine the challenges this presents to OEMs and review the options available to them including how open initiatives like ONL, SAI and TAI are making this easier. Finally, we will present an architecture that will allow an OEM to create a portable NOS, based on ONL, with fast time to market, a high degree of flexibility and with the potential to be deployed in the access, aggregation and core networks not currently targeted by any disaggregated, portable, NOS.

Speakers
avatar for Paul Carter

Paul Carter

Director of Product Management, Metaswitch Networks


Friday March 15, 2019 8:00am - 8:25am PDT
210 DH

8:00am PDT

OCP NIC Technical and Ecosystem Update
OCP NIC form factor is becoming more popular in the past 5 years. 2019 is an important year of this community project: OCP NIC 3.0 is getting more mature; active design-in and validation is happening with NIC suppliers, system suppliers, and hyperscale users.Join us at 2019 OCP Summit to learn the latest progress across technical topics, and ecosystem status.

Speakers
avatar for Joshua Held

Joshua Held

Mechanical Engineering Manager, Facebook
avatar for Jia Ning

Jia Ning

Hardware Engineering Manager, Facebook


Friday March 15, 2019 8:00am - 8:25am PDT
210 A

8:00am PDT

Open Systems Firmware Overview
Open Systems Firmware is growing fast. From zero systems in 2017 we expect to see 100,000 or more deployed in 2019 in switches, servers, and other types of systems. OSF is also a key part of partner efforts to support a circular economy model in which systems are reused, not sent to landfills. In this session, we'll go over the state of the effort and where we're headed.

Speakers
GG

Gundrala Goud

Microsoft
Gundrala ""Devender"" Goud is currently an Engineering Director at Microsoft, Redmond campus, and he works in the Cloud Hardware Solutions development group. Prior to joining Microsoft, Devender worked at Intel in the Data Center Group for 17+ years and held various technical leadership... Read More →
avatar for David Hendricks

David Hendricks

Firmware Engineer, Facebook
David got his first taste of BIOS tweaking as a kid overclocking gaming PCs and was drawn into the coreboot world where he learned real firmware hacking. Since then he has developed, deployed, and advocated for open source firmware in consumer electronics and datacenter infrastructure... Read More →
avatar for Ron Minnich

Ron Minnich

Software Engineer, Google
linuxboot, u-root, coreboot, linuxbios, ... all open source firmwarelinux kernel, servers,


Friday March 15, 2019 8:00am - 8:25am PDT
Marriott Salon III

8:00am PDT

PCI Express: Enabling Performance in HPC
PCI Express® (PCIe®) architecture has enabled storage and data centers to transition to high speed, low latency power efficient performance over the past few years. However, the demand for additional performance in power constrained devices and data-hungry server, client, embedded, IoT and mobile market segments is growing. To meet the need of these data demanding markets, PCI-SIG® continues its nearly three decade history of delivering performance doubling with the development of the PCIe 4.0 and PCIe 5.0 specifications. In this sessions, attendees will: • Gain insight into the first FYI testing for PCIe 4.0 specification and product rollout • Receive a firsthand look at the PCIe 5.0 spec delivering 32 GT/s targeted for completion in Q1 2019 • Understand PCIe low power capabilities and options • Review case studies where PCIe is being deployed in OCP use cases

Speakers
RS

Richard Solomon

PCI-SIG Vice-President and Co-chair Compliance Workgroup, Synopsys
Richard Solomon is the Technical Marketing Manager for Synopsys' DesignWare PCI Express Controller IP.  He has been involved in the development of PCI chips dating back to pre-1.0 versions of the PCI spec. Richard architected and led the development of the PCI Express and PCI-X interface... Read More →


Friday March 15, 2019 8:00am - 8:25am PDT
210 CG

8:00am PDT

SAI - Updates and Roadmap
In 2018, the SAI community has advanced the standard intensively in RDAM and Telemetry area. We will update the industry on how SAI evolved as well as where it is going.

Speakers
avatar for Jai Kumar

Jai Kumar

Distinguished Engineer, Broadcom
Experienced Distinguished Engineer with a demonstrated history of working in the networking and semiconductors industry. Skilled in L2/L3 forwarding technologies, Network Processors, Data Center, Network Design, and Network Data Analytics. Strong engineering professional with 18 patents... Read More →
GL

Guohan Lu

PRINCIPAL SE Manager, Microsoft


Friday March 15, 2019 8:00am - 8:25am PDT
Marriott Salon IV

8:00am PDT

Telco Project Welcome & Status
Speakers
avatar for Craig White

Craig White

Product Architect, Airframe R&D, Nokia
I have been in Open compute Telecom project from its creation. I was fortunate to have been Telecom project co-leader with Bill Carter. I work at Nokia in Mika Hattanpaa\\'s Architecture R&D tribe lead by Samuli. I was responsible for Open Compute Foundation work and am currently... Read More →
avatar for Sumithra Bhojan

Sumithra Bhojan

PMTS, AT&T
Sumithra Bhojan is a Principal Member of Technical Staff at AT&T with a focus on Technology Design and Architecture for Broadband Wireline Access systems and RAN Fronthaul Transport. In her current role in AT&T labs, she has been focused on AT&T’s vision of transitioning access... Read More →


Friday March 15, 2019 8:00am - 8:25am PDT
212 CD

8:00am PDT

Open Source Firmware Meetup
Firmware Developers' Community Meetup

Speakers
avatar for Sai Dasari

Sai Dasari

Software Engineering Manager, Facebook


Friday March 15, 2019 8:00am - 5:00pm PDT
Marriott Salon V&VI
  EW: OSF/Security
  • YouTube Link -

8:20am PDT

Google 48V Rack Adaptation and Onboard Power Technology Update
Overview As disclosed in OCP 2016 keynote, Google rack power utilizes direct current 48V. In 2017, Google contributed the Open Rack V2 48V spec to help the development of 48V rack and power components. In 2018 OCP, to further develop the 48V ecosystem, Google shared project “Flatbed” which is an adapter board and specifications to integrate 12V motherboards to a 48V rack. We also shared a new on-board 2-stage 48V to PoL voltage conversion approach to improve Total cost of ownership (TCO). For 2019 OCP, we like to present the toolkit to integrate AC-based chassis into 48V rack and new progress for on-board 48V power conversion. AC-12V based PSU to DC/DC PDB Approach The OTS (Off The Shelf or enterprise) payloads are designed to take AC as input with the traditional 2 conversion UPS in the data center level, and it often requires certain hold up time (15ms to 20ms) and N+N redundancy to meet the availability target and repair process requirement. To optimize the power delivery system, many data center consumers have already moved to a centralized rack power delivery architecture which have the battery in the rack or tray level. The challenges of accommodating the OTS payloads into centralized rack power delivery system become very obvious because of the power architecture difference. In this workshop, we are going to show one approach to smoothly integrate the AC-based chassis OTS payloads into 48V centralized power system to meet a critical TTM target with a simple toolkit. This approach has been enabled by simplifying the power requirements in hold up time, redundancy, narrow input range, modularized design and so on. Onboard power In terms of 48V onboard power delivery, Google has been continuously driving the technologies toward higher efficiency, higher density, better scalability and cost effectiveness to meet the rapidly increasing data center computing power demand. Our technologies address a variety of use cases across server, storage, networking and AI accelerators. Significant progress has been made in these specific applications: 1. multi-phase 48V-to-12V voltage regulators adopting the latest hybrid switched-capacitor-buck topologies for traditional 12V workloads such as PCIEs and OTS servers; 2. very high efficiency high density fixed ratio bus converters for 2-stage 48V-to-PoL power conversions; 3. high frequency high density voltage regulators for extremely power hungry AI accelerators. A complete onboard 48V power solution portfolio is explored in this workshop.

Speakers
SJ

Shuai Jiang

STAFF POWER ENGINEER, Google
XL

Xin Li

Technical Lead Manager at Google Platforms, Google


Friday March 15, 2019 8:20am - 8:38am PDT
210 BF

8:20am PDT

An Update on the New SNIA Computational Storage Technical Work Group: Contributing to Industry Understanding and Use of This New Paradigm of Computer Architecture
The industry is seeing an increase in customer requirements to move compute closer to traditional storage devices and systems. In response, a growing number of data-driven applications have demonstrated that adding computation to the normal storage features of devices and systems can realize a significant performance and infrastructure scaling advantage. Computational Storage solutions typically target applications where the demand to process ever-growing storage workloads is outpacing traditional compute server architectures. These applications include AI, big data, content delivery, database, machine learning and many others that are used industry-wide. SNIA has established a new Technical Work Group to facilitate the user of computational storage in mainstream application environments. This session will discuss their work activities and plans to promote interoperability of devices and standards for system deployment, provisioning, management and security.

Speakers
avatar for Tom Friend

Tom Friend

CEO/CBW, Illuminosi Inc.
Helping the world to build better storage, I am excited about SNIA's Computational Storage project and the potential improvements it brings to Big Data.
avatar for Scott Shadley

Scott Shadley

Storage Technologist, VP Marketing, Co-Chair SNIA Computational Storage TWG, NGD Systems
Doing my best to improve the world with NVMe Computational Storage Solutions from NGD Systems. Lover of Technology, Innovation and a bit of a Socialite.If it is Storage, I dabble...


Friday March 15, 2019 8:20am - 8:38am PDT
212 AB

8:30am PDT

A Large Scale Deployment Experience Using Immersion Cooling in Datacenter
With the improvement of computing performance, the power consumption is significantly increased. Increasing power consumption of server components brings enormous challenges to heat dissipation and power optimization. Immersion cooling completely breaking the air-cooling power dissipation bottleneck and also achieving extremely low PUE. This presentation introduces the problems encountered by Alibaba in the scale deployment of immersion cooling data center, solutions and considerations.

Speakers
avatar for Yangfan Zhong

Yangfan Zhong

Senior Engineer, Alibaba


Friday March 15, 2019 8:30am - 8:55am PDT
210 E

8:30am PDT

OCP Ready Facility Recognition Program for Colo Solution Providers
The OCP Ready Facility Recognition Program for Colo Solution Providers was developed to bring together facility providers and end-users of OCP hardware. The unique designs developed in the pursuit for efficiency and scalability can be a challenge to deploy in facilities that are not capable. This session is aimed at facility providers with interest in learning about the program and end-users in understanding the benefit of deploying into an OCP Ready Facility.

The session will include an introduction to and testimonial from Amsterdam colocation provider, Switch Datacenters.

Speakers
avatar for Edgar van Essen

Edgar van Essen

Managing Director, Switch Datacenters
Edgar Van Essen has over 20 years of general and commercial leadership (P&L) in high tech, global companies in the EMEA region. Successful consistent track record in leading structural business change and exponential growth in renowned high tech and innovative companies. Before a.o... Read More →
avatar for Brevan Reyher

Brevan Reyher

Design & Build Engineer/Data Center Facility Project Lead, Rackspace/OCP Foundation
Brevan Reyher leads the Global Data Center Infrastructure Design & Build DC Engineering practice area at Rackspace Hosting. After graduating from Trinity University in San Antonio, he has spent over 10 years in the design, construction, operation and maintenance of data center facilities... Read More →


Friday March 15, 2019 8:30am - 8:55am PDT
211 AB

8:30am PDT

Modeling Immersion Cooling Compatible with OCP Profile
In this workshop, we will present the design and implementation of Wiwynn rack-scale 2-phase immersion cooling management solution. First, we will examine the data schema which defines logical and physical objects of a tank conforms to applicable OCP Profiles. Then we will explain the additional attributes and actions with examples of the adopted chassis types needed for managing a tank. At a quick glance, a tank looks like a super huge enclosure with a bunch of semi-enclosed sleds. However, some intermediate layers in the object hierarchy are needed to signify the influences between each other. For example, we adopt the ChassisType “Cartridge” to contain two bound sleds that cannot be plugged in to or removed from a tank independently. On the other hand, we also implement specific immersion cooling tank management mechanism. For example, we conduct power capping or workload migration to suppress vapor escape while opening a tank lid. At last we will give a video demonstration of the management for Wiwynn Immersion cooling tank.

Speakers
SY

Shao Yen

Senior Engineer, Wiwynn Corporation


Friday March 15, 2019 8:30am - 8:55am PDT
Marriott Salon I&II

8:30am PDT

The New Facebook DC Topology
Facebook has a new fabric design to accommodate the ever-growing demand of applications. The new design is both simpler and more scalable. This will cover the reasons for the new design and how we handled the migration from our previous fabric design to this design (which is one of our most time-consuming steps in deployment of new topologies).

Speakers
AA

Alexey Andreyev

Facebook Network Team, Facebook


Friday March 15, 2019 8:30am - 8:55am PDT
211 CD

8:30am PDT

ONIE Project Update
A brief overview what ONIE is, followed by an update on the state of the project, and a listing of some future goals

Speakers
AD

Alex Doyle

ONIE Project Lead / Build engineer, Cumulus Networks
I'm the project lead for the Open Network Install Environment


Friday March 15, 2019 8:30am - 8:55am PDT
210 DH

8:30am PDT

The Journey of OCP NIC3.0 Solution Integration and Demo
Facebook has been zealous of OCP NIC since its introduction in 2012. Fast forwarding 3 generations later in year 2019, now we have OCP NIC3.0. The new multi-host NIC's lead development is a co-operation with prominent NIC suppliers on our flexible YosemiteV2 platform. OCP NIC3.0 enhanced multi-host out-of-band (OOB) management flexibility with independent power control and greater sensor monitoring through PLDM over NCSI. We are pleased to share our learnings on solution integration, OOB management and initial bring-up to the community.

Speakers
avatar for Damien Chong

Damien Chong

Hardware Engineer, Facebook
BW

Ben Wei

System Software Engineer, Facebook


Friday March 15, 2019 8:30am - 8:55am PDT
210 A

8:30am PDT

Case Study: Alternatives for SMM usage in Intel Platforms
The broadcast System Management Mode (SMM) model has been used for many years to manage priority system events but has a number of disadvantages. Overuse of System Management Interrupts (SMI) results in performance degradation, increases latency with higher core counts, and introduces potential race conditions. SMM is also difficult to debug and has access to system resources outside of the OS environment, which makes it target for firmware exploits. This session expands on Intel’s initiative to reduce SMM footprint and provide alternatives for handling runtime platform events. Intel described SMI reduction methods based on Protected Runtime Mechanism (PRM), UEFI Capsule, and the Baseboard Management Controller (BMC) at the 2018 OCP Regional Summit. The presentation features a case study and demonstration using Intel® Xeon® Scalable Processors with EDK II firmware.

Speakers
avatar for Sarathy Jayakumar

Sarathy Jayakumar

Principal Engineer, Intel Corp


Friday March 15, 2019 8:30am - 8:55am PDT
Marriott Salon III

8:30am PDT

Transponder Abstraction Interface (TAI)
In this session, I will introduce the Transponder Abstraction Interface (TAI, https://github.com/Telecominfraproject/oopt-tai). TAI defines the API to provide a vendor-independent way of controlling transponders from various vendors. It is heavily inspired by OCP SAI. Together with SAI, TAI can enable open transport ecosystem.

Speakers
avatar for Scott Emery

Scott Emery

Principle Engineer, Cumulus Networks
I'm the OCP Networking Project Co-Lead and an Engineer at Cumulus Networks. I'm interested in pretty much anything related to networking, and especially open networking. Or, the San Jose Sharks.
avatar for Wataru Ishida

Wataru Ishida

Field Application Engineer, NTT Electronics America


Friday March 15, 2019 8:30am - 8:55am PDT
Marriott Salon IV

8:30am PDT

Cost-effective Approach for Telco Network Analysis in 5G
TINA is a unified network analytics platform to guarantee quality of SKT's mobile services running on Telco networks. But, in 5G we must achieve this mission in cost-effective way. To this end, we leverage OCP powered networking whitebox with Programmable Ethernet Switching Chip and Open Source software. In this session, we will share our lesson learned on Programmable Ethernet Switching Chip and Open Source Big Data and Machine Learning software stack to handle tens of millions of packets per second per server node.

Speakers
avatar for Junho Suh

Junho Suh

Research Engineer, SK Telecom
Currently, I am a research engineer at SKT. Before joining SKT, I received a Ph.D. in network systems at Seoul National University in 2015. Prior to SNU, I received BS at KAIST in Electrical Engineering and Computer Science in 2008. While doing BS, I joined WINE lab and carried out... Read More →


Friday March 15, 2019 8:30am - 8:55am PDT
212 CD

8:30am PDT

Open Domain-Specific Architecture (ODSA) Sub-project Launch
Today there are many companies affected by, and affecting SoCs and domain-specific accelerators, such as co-processors for networking, security, storage, machine learning and inferencing.

This workgroup will define an architecture that enables the mixing and matching of available silicon die from different suppliers onto a single SoC for data center applications. The goal is to define a process to integrate best-of-breed chiplets onto an SoC.


Speakers
avatar for Bapi Vinnakota

Bapi Vinnakota

Engineer, Netronome
Bapi is the Director of Silicon Architecture Program Management and OCP ODSA Workgroup Lead at Netronome. Bapi taught at the University of Minnesota, Twin Cities, immediately after a Ph.D. at Princeton. He joined Intel through an acquisition and was the architect of a flow processor... Read More →


Friday March 15, 2019 8:30am - 9:00am PDT
210 CG

8:40am PDT

SSD with Compression for the Compute and Storage Infrastructure: Implementation, Interface and Use Cases
Data compression within the SSD significantly reduces write amplification leading to increased performance and endurance. Alternatively, compression inside the SSD can be utilized to significantly increase user capacity. These benefits can in particular make high capacity QLC SSDs more attractive for the data center, cloud or edge infrastructure. However, compression inside SSDs is not widely available yet, and there is a lack of understanding of the implementation trade-offs and benefits. This talk will describe the implementation, interface and use cases for compression inside SSDs. The goal of this presentation is to add compression-enabled SSD devices to the Open Compute Project by defining hardware and software interfaces, workloads, and use cases. This will provide the industry an additional attribute to consider when optimizing SSD requirements for specific workloads.

Speakers
EH

Erich Haratsch

Managing Technologist, Seagate Technology


Friday March 15, 2019 8:40am - 8:58am PDT
212 AB

8:40am PDT

48V 2-stage System Efficiency Optimization by using STC Converter with Dynamic Converting Ratio
Google’s proprietary STC employs LC resonant tank to realize soft charging and soft switching. This technology enables high efficiency high density 48V bus as the first stage conversion in a two stage step down system. Two stage combined efficiency varies in various load conditions. High STC converting ratio provides lower input voltage of second stage which leads to less switching loss in 2nd stage VRs. Lower converting ratio gives less current stress of 1st STC converter in high output power range. As a result, high converting ratio is beneficial in lower output power application, while low converting ratio has advantage in higher output power. Wiwynn is proposing a dynamic converting ratio STC design to further optimize 2-stage combined efficiency by altering converting ratio based on output power. • High converting ratio operation in lower load • Parallel resonant legs and operates in low converting ratio in heavy load

Speakers
SY

Sam Yang

Associate Technical Director, Wiwynn Corporation


Friday March 15, 2019 8:40am - 9:05am PDT
210 BF

9:00am PDT

OCP NIC 3.0 Design and Implementation Experiences
OCP NIC 3.0 specification as a follow-on to OCP Mezz 2.0 specification has enabled a wider spectrum of next generation NIC board designs. As a leading provider of NICs and Smart NICs, Broadcom has been at the forefront of OCP NIC 3.0 specification and designs. In this talk, we will share Broadcom’s implementation experiences with OCP NIC 3.0 designs. We will cover mechanical, electrical, thermal, host interface, manageability, and firmware security aspects of OCP NIC 3.0 designs.

Speakers
LN

Long Nguyen

Master Hardware Engineer, Broadcom Inc.
avatar for Hemal Shah

Hemal Shah

Distinguished Engineer, Broadcom
Hemal Shah is a Distinguished Engineer and Systems/Software/Standards architect in the Compute and Connectivity (CCX) division at Broadcom Inc. He leads a team of architects. In his role, he is responsible for the product architecture definition of Ethernet NIC products, NIC integration... Read More →


Friday March 15, 2019 9:00am - 9:12am PDT
210 A

9:00am PDT

Advanced Network Telemetry and Analytics Over SONiC
The competition in the cloud market is driving innovation in the modern data center. To maintain a competitive advantage, companies are constantly looking for ways to provide end Consumers with more choice and more innovation. For cloud providers, these business drivers also require continuous optimization to run leaner, raise efficiency and reduce CapEx and OpEx. OCP SONiC and SAI projects are some of the key innovation drivers for achieving these goals. Data center, who are at the leading edge of this paradigm shift, are building highly scalable groundbreaking solutions based on SONIC and SAI leveraging cutting edge features offered by merchant silicon. The Open source BroadView™ project perfectly complements SONiC and SAI initiatives by delivering a consistent northbound interface to network monitoring and collector applications. Built with a flexible architecture and uses a plug-in model to interact with SAI on the southbound side, the Broadview agent is a platform agnostic open source software module that exports open REST API to applications further simplifying development and deployment of advanced services in next-generation networks. In this presentation, we will outline how SONiC customers can leverage Broadview Agent which is compliant with the SAI-TAM 2.0 spec, and integrated with SAI, to deploy many unique and advance telemetry features.

Speakers
avatar for Eli Karpilovski

Eli Karpilovski

Director of Marketing, Broadcom Inc.
Eli manages the software product management and marketing team for the Broadcom switch business unit: SDK, drivers, embedded , telemetry and programmability applications, and 1588 based solutions. Follow him on Twitter: @Elikarpilovski.


Friday March 15, 2019 9:00am - 9:12am PDT
Marriott Salon IV

9:00am PDT

Xhaul Switch Specification
New OCP specification of a xhaul Switch used in mobility network using merchant silicon.

Speakers
MM

Mike Meche

Principle Member of Tech Staff, AT&T Converged Access & Device Technology


Friday March 15, 2019 9:00am - 9:18am PDT
212 CD

9:00am PDT

Unleash Stranded Flash Capacity - Deep Dive on Disaggregated Storage Architecture, Trends and OCP Solutions
Flash memory is becoming an increasingly popular choice for latency sensitive workloads such as databases, real time analytics, in-memory key-value stores. Current generation of NVMe SSDs provide significant performance at a lower latency. It is not uncommon to expect these devices providing million IOPS. In traditional deployments with NVMe SSDs directly attached to the processor, it is difficult to achieve balanced system configuration resulting in stranded flash performance and capacity. In disaggregated storage, flash is shared among pool of compute nodes using dynamic binding of storage resources based on performance, capacity requirements. However, carving out storage from the flash pool and exposing over network introduces latency. In this session we outline different network protocols, latency implications and industry trends to deliver low latency disaggregated flash storage over the network using OCP platforms.

Speakers
avatar for Anjaneya ‘Reddy’ Chagam

Anjaneya ‘Reddy’ Chagam

Senior Principal Engineer and Lead Cloud Storage Architec, Intel
Anjaneya “Reddy” Chagam is a Senior Principal Engineer and Lead Cloud Storage Architect in Intel’s Cloud and Enterprise Solutions Group. He is responsible for developing software-defined storage strategy, architecture, and technology initiatives. He was a member of the Technical... Read More →
avatar for Manoj Wadekar

Manoj Wadekar

Storage Hardware Architect, Facebook
Current Focus: Solve ambiguous problems for hyper-scale infrastructure. Find the problems, drive innovative solutions.Background:28+ years in Server, Storage, and Networking industry.Specialties: Transforming hyper-scale infrastructure to drive performance and efficiency (proven for... Read More →


Friday March 15, 2019 9:00am - 9:20am PDT
212 AB

9:00am PDT

Introduction to ACS Immersion Standards and Best Practices
Presenting the OCP AC Immersion workstream progress and everything you need to know before working with or for immersion technology.

Speakers
avatar for Rolf Brink

Rolf Brink

CEO, Asperitas
Product development for cloud architectures and datacentre infrastructure, combined with a passion for cleantech innovations like liquid cooling, are the foundation of the developments which drive Asperitas.Asperitas is a clean-tech company focused on greening the datacentre industry... Read More →



Friday March 15, 2019 9:00am - 9:25am PDT
210 E

9:00am PDT

Tools and Process for Creating a Redfish Profile
This session will describe the machine-readable document format for describing a Redfish Interoperability Profile, and will discuss the open source tools available to generate documentation and test systems for conformance.

Speakers
JA

Jeff Autor

Distinguished Technologist, DMTF


Friday March 15, 2019 9:00am - 9:25am PDT
Marriott Salon I&II

9:00am PDT

Hardware for the New Facebook DC Topology
Speakers
SK

Sami Khan

Technical Program Manager, Facebook
avatar for Hao Shen

Hao Shen

Hardware Engineer, facebook
Hao is hardware engineer at Facebook working on AI platform. His focus is on AI Hardware System, Accelerator enablement and large scale datacenter solution.
XW

Xu Wang

Hardware Engineer, Network, Facebook


Friday March 15, 2019 9:00am - 9:25am PDT
211 CD

9:00am PDT

Data Processing Acceleration Over OCP Based Solutions
OCP has been a pioneer in both creating open hardware and software standards that promote multi-vendor collaboration and create an ecosystem that benefits both producers and consumers of technology. As Moore’s law slows down the need becomes more critical for acceleration technologies that offload specific data management functions including packet processing and security management. As each vendor may have different implementations of these acceleration co-processors, it is critical to innovate open software interfaces that are hardware-independent and are developed in the open-source community. Importantly the same software APIs should work transparently, whether the functionality is implemented in pure software or using hardware to accelerate data processing. This session will focus on new, open interfaces being developed upstream in the kernel and available to support OCP hardware that accelerate virtual switching, quality of service, traffic class management, security, and storage.

Speakers
avatar for Dror Goldenberg

Dror Goldenberg

SVP SW Arch, Mellanox Technologies
Dror joined Mellanox as an architect in 2000 to work on exciting network innovations. Dror drove silicon and system architecture of multiple generations of NICs, Switches and SoCs. Dror’s main focus nowadays is on software architecture, enabling network accelerations of cool technologies... Read More →


Friday March 15, 2019 9:00am - 9:25am PDT
210 DH

9:00am PDT

Exploiting the True Potential of At-Scale Debug
As we move to cloud computing and the need for more compute power closer to data at the edge grows over time At-Scale debug provides an opportunity to mine telemetry like never before. While the channel has been opened up there are several considerations into what specific underlying telemetry is key at what point in time and what is the best way to collect it, visualize it and use it for decision making to improve the overall Datacenter posture. Is telemetry really limited to crash dumps ? No its much more than that and that is what we want to share through this joint Intel and Lenovo presentation.

Speakers
GG

Gundrala Goud

Microsoft
Gundrala ""Devender"" Goud is currently an Engineering Director at Microsoft, Redmond campus, and he works in the Cloud Hardware Solutions development group. Prior to joining Microsoft, Devender worked at Intel in the Data Center Group for 17+ years and held various technical leadership... Read More →
SK

Sumeet Kochar

Chief Architect Firmware Lenovo DataCenter Group, Lenovo
20+ years of low to highend server firmware and system management experience with expertise around memory, chipset and RAS.
TY

Theodros Yigzaw

Platform RAS Architect, Intel
RAS Architecture. Or anything else, really.


Friday March 15, 2019 9:00am - 9:25am PDT
Marriott Salon III

9:00am PDT

Standardizing the Modular Data Center Specification
This session will be jointly hosted by Swedish Modules and Schneider Electric. We request a full hour and the session will cover: 1) The work done by the MDC sub project to create MDC specifications and the intended benefits tot he community. 2) Swedish Modules presents the details of their 300kW multi Module design 3) Schneider Electric presents the details of their 90kW all-in-one design

Speakers
avatar for Scott Neal

Scott Neal

Product Director - Prefabricated Modules, Schneider Electric
I spend most of my day developing, evolving, and promoting modular data centers for Schneider Electric. My 20+ year history in the data center industry has been in a mix of engineering, sales, and product development roles. What I enjoy most is solving customers most critical c... Read More →
avatar for Roberto Söderhäll

Roberto Söderhäll

Chief Business Development Officer, Swedish Modules
Roberto is a veteran in working with innovation within the Data Center Industry. Spending more than two decades in the Data Center Industry, Roberto has been challenging the Status Quo of the Data Center Industry. As a speaker in many events his been proclaiming the Data Center as... Read More →


Friday March 15, 2019 9:00am - 9:50am PDT
211 AB

9:10am PDT

Flex Furthers Direct Conversion Ecosystem
The renewed interest in Direct Conversion, using isolated DC-DC converters to convert power from a 48V bus in data centers directly down to voltages suitable for powering processors, FPGAs and ASICs, can be attributed largely to energy savings. In December 2017, it was reported that data centers consume about 416 terawatts globally (Forbes) and, with the explosion in cloud computing, the figure is surely higher now and set to increase dramatically. Small percentage savings are therefore valuable in terms of cost reduction and reduced environmental impact. Today’s power distribution architectures typically convert either AC or 48V DC to an isolated 12V, which is routed around server racks with non-isolated point-of-load converters providing the final step-down in voltage. As the rack loading has ramped up, however, the current at 12V has become increasingly unmanageable, causing significant ohmic distribution losses. Dual conversion of 48V to 12V, then 12V to load voltage, has also been a source of losses, even with end-to-end efficiencies of about 89%. A solution already adopted by Google and Facebook in their standard Open Rack v2.0 is to eliminate the 12V bus for the very highest current demands and convert 48V directly to the load voltage – the Direct Conversion approach. In this arrangement, 48V is distributed directly to the server blades, and an isolated converter on the board close to the load provides precise regulation where it is needed. This approach had been adopted in the past, when end voltages were typically 5V or 3.3V, but the conversion technology of the day was too inefficient at voltages lower than these. Technology has now advanced, to the point where an isolated 48V to 1V converter can be more than 92% efficient: a 2% to 3% gain over the IBA approach. Benefits are wide-ranging: current at 48V is a quarter of the 12V value for the same power, yielding a sixteenth of the losses in comparable cabling and bus connections; overall footprint is smaller, with a single conversion stage, and hardware costs lower; cooling requirements are reduced, and energy saved with a more efficient process. The modern design of the new Direct Conversion DC-DCs also facilitates extra functionality such as Adaptive Voltage Scaling (AVS) for optimal processor speed and energy draw, and a serial voltage ID interface. Full digital control of the converters is typical, with programmability through an I2C PMBus interface. The first products from Flex Power Modules, based around ST microcontrollers, provide a modular multi-phase solution. There is a 70A “main” unit with a footprint of 28mm x 12mm, while up to five additional 70A “satellites” in the same footprint can be added as needed, up to 420A in total. This approach offers efficiency optimization across a wide range of current demands, with the addition of automatic phase addition and removal where demanded by the load. With a 1.0V output, a nearly flat efficiency of 92% is provided across most of the load curve. Even with all the advantages Direct Conversion affords, the marketplace insists on having a broad base of suppliers. To that end, Flex Power Modules is expanding the 48V ecosystem for Direct Conversion. Now that Flex Power Modules has introduced products, and continues with a product roadmap, the ecosystem is here and growing. Flex is one of the founders of an industry alliance, the Power Stamp Alliance (PSA), further promoting the expansion of the 48V ecosystem. The PSA is a collaboration between founding members Flex Power Modules, Artesyn Embedded Technologies, Bel Power Solutions and STMicroelectronics to offer Direct Conversion products. It sets functional and mechanical standards for these new Direct Conversion products, with each member free to set the topology utilized and detail design. This presentation will highlight the key technical features and benefits of the 48V Direct Conversion approach in OCP platforms, as well as highlighting the commercial benefits of an open alliance such as the PSA.

Speakers
avatar for Bob Cantrell

Bob Cantrell

Senior Field Engineer, Flex Power Modules
Bob Cantrell is a Senior Field Application Engineer for Flex Power Modules. He has over 30 years of experience working in the power industry.  Prior to working at Flex, Bob was a Senior Application Engineer at Ericsson Power Modules, a Member of Technical Staff at Lucent Technologies... Read More →


Friday March 15, 2019 9:10am - 9:30am PDT
210 BF

9:15am PDT

Multi-mode OCP NIC 3.0 card HW management
The introduction of OCP NIC 3.0 brings new HW management challenges. OCP NIC 3.0 standard allows for multiple operating modes for NIC cards. Multi-Host capable OCP NIC 3.0 cards based on ConnectX, supporting both Mellanox Multi-Host® and Mellanox Socket Direct® technologies, leverage this new standard by offering a single hardware, capable to operate as Single-Host, Multi-Host and Mellanox Socket Direct® device. Being an OCP NIC 3.0 card, the operating-mode is automatically set, allowing data-centers and OEMs to use the same hardware in different server platforms. This talk covers the special features which were introduced into this new technology, to address the new HW management challenges.

Speakers
YI

Yuval Itkin

Distinguished architect, Mellanox
Servers management and OCP server hardware.


Friday March 15, 2019 9:15am - 9:27am PDT
210 A

9:15am PDT

Cloud Scale Telemetry – Real-time Telemetry and Analytics at Scale
The shift to more efficient cloud-based provisioning requires new diagnostic and operational characteristics for the data center network. The operation team’s knowledge of where a server was and what it was doing started to alter with early virtualization technologies and will change entirely as the use of Linux containers continues to expand. The need to understand far more about the state of the infrastructure and the state of the applications running on the infrastructure requires more telemetry information than has traditionally been available. Cisco Cloud Scale ASICs support a number of hardware-based sources of analytic information, and that includes the Streaming Statistic Telemetry (SSX) capability, which streams ASIC state data in an open format. State data for example includes interface statistics, buffer and queue information, and control plane changes. In this session, we will talk about the various hardware telemetry capabilities in Cisco Cloud Scale ASICs, and telemetry and analytics solutions that we have available to visualize and operationalize around the telemetry data. We will also talk about how we support open community initiatives in Data-Plane Telemetry and how we expose the SSX capability to SONiC, as well as conduct a live demo for the audience to see how SSX works in action!

Speakers
WC

William Chen

Product Manager, Cisco
AK

Ashoka Kallappa

Principle Engineer, Cisco
TS

Tim Stevenson

Distinguished Engineer, Technical Marketing, Cisco


Friday March 15, 2019 9:15am - 9:27am PDT
Marriott Salon IV

9:19am PDT

AT&T Spec Update
Specification of Whitebox design for converged access device to provide PON and xhaul function using FPGA

Speakers
avatar for Sumithra Bhojan

Sumithra Bhojan

PMTS, AT&T
Sumithra Bhojan is a Principal Member of Technical Staff at AT&T with a focus on Technology Design and Architecture for Broadband Wireline Access systems and RAN Fronthaul Transport. In her current role in AT&T labs, she has been focused on AT&T’s vision of transitioning access... Read More →


Friday March 15, 2019 9:19am - 9:37am PDT
212 CD

9:25am PDT

OCP Use Case: Using a Software Defined Storage on OCP Lightning (NVMe JBOF)
Software Defined Storages is more flexible and scalable compared with Storage Appliances, because it can run on commodity hardware. OCP Lightning (NVMe JBOF) is suitable for these applications. Yahoo Japan is starting to use Quobyte that is software storage as backend of OpenStack Cinder, Manila, Kubernetes Persistent Volumes. In this talk, we will describe about our experiences of using a Software Defined Storage on OCP Lightning.

Speakers
avatar for Akira Kamio

Akira Kamio

CyberAgent, Inc.
Akira Kamio is a Software Engineer in CyberAgent's Private Cloud team. CyberAgent provides Internet advertising, medias, games in Japan.He mainly design, development, verification and operation of openstack-based private cloud in CyberAgent. He likes hypervisor optimization, such... Read More →
avatar for Yusuke Sato

Yusuke Sato

Manager, Yahoo Japan
Yusuke Sato is responsible for private cloud compute and storage at Yahoo Japan Corporation. He has been engaged mainly in verification of server hardware, operation of storage systems, and establishment and operation of virtual environments. He led the server and OS team from 2013... Read More →


Friday March 15, 2019 9:25am - 9:45am PDT
212 AB

9:30am PDT

NIC Safe Mode
A key element in cost reduction of OCP based platforms, relies on the minimization of connections on each server. When miss-configuration prevents a server from starting-up, it can mandate physical access to the machine to ratify the configuration, resulting in increased down-times and operational cost. NIC Safe-Mode technology allows to eliminate the need for physical access to the server. With NC Safe-Mode technology, the NIC automatically detects the problematic configuration scenario, changes its settings into safe-mode settings, allowing to remotely ratify the incorrect settings and thus lowering the operational costs of such incident.

Speakers
YI

Yuval Itkin

Distinguished architect, Mellanox
Servers management and OCP server hardware.


Friday March 15, 2019 9:30am - 9:42am PDT
210 A

9:30am PDT

InBand Network Telemetry an Open and Multi-platform Network Analytics for Disaggregated Data-Centers
Modern data-centers use high-density L2 and L3 topologies to accommodate heterogeneous mission critical applications. In such an environment visibility is essential to reliably operate and manage the networking elements. We present an open framework called “Inband Network Telemetry” (INT). In INT each network element (e.g. a SmartNIC or a Barefoot Network switch) can be programmed to insert metadata about the network state in the packet as it traverses the network. This enables visibility on all traffic (e.g. Compute, Storage/RDMA) carried over ethernet for transient issues due to bottlenecks, failures or configuration errors. In such a solution machine learning techniques can be used to Baseline Network Performance (Path and Latency), match patterns of behavior and correlate phenomena. Such techniques can be used to promptly react to network issues or predict and anticipate problems.

Speakers
avatar for Roberto Mari

Roberto Mari

DIRECTOR BAREFOOT NETWORKS, Barefoot Networks
Roberto spent 16 years in the IT industry, covering Data Center Routing, Switching and WAN optimization technologies.He was one of the early leaders in the Product Team that built and launched the VMware NSX Network and Security virtualization platform to market. Before that he was... Read More →


Friday March 15, 2019 9:30am - 9:42am PDT
Marriott Salon IV

9:30am PDT

Liquid Cooling Trends and Requirements for Racks and Servers
Liquid cooling technology is an one of the pillars for unlocking next generation HW. The term liquid cooling can mean many things to many people, this presentation discusses an overview of the technology from a cloud point of view showing MSFT Olympus server ability to adapt to different liquid cooling technologies.
It will also discuss the need for supporting liquid cooling technology and its supply chain.

Speakers
avatar for Husam Alissa

Husam Alissa

Principal HW Engineer, Microsoft
Husam is a technical lead working for Microsoft Data Center Advanced Development Team with focus on next generation technologies for hardware and infrastructure.
avatar for Brandon Rubenstein

Brandon Rubenstein

Principal Platform Engineer, Microsoft
I've had the privilege to contribute to Microsoft's success as an engineering director, a design architect and a lead designer. Most recently I've managed the mechanical and power teams responsible for defining, designing, and maintaining Microsoft's cloud business hardware and datacenter... Read More →


Friday March 15, 2019 9:30am - 9:55am PDT
210 E

9:30am PDT

OpenBMC - Project Update
An update for the OCP community on what has happened in the last year in the OpenBMC project in terms of technical content, and community structure.

Speakers
avatar for Brad Bishop

Brad Bishop

Software Engineer, IBM
Brad Bishop is a software engineer at IBM. He has been a member of IBM's BMC software stack development organization since 2001. More recently he was a core contributor to the inaugural appearance of OpenBMC on the Barreleye platform and was appointed to the technical steering committee... Read More →


Friday March 15, 2019 9:30am - 9:55am PDT
Marriott Salon I&II

9:30am PDT

9:30am PDT

Large-scale Operations for our Next-gen Fabric and Fabric Hardware
There is a lot of testing to get ready for a new fabric and new hardware, including large-scale device and network testing, as well as operational migration, configuration, and image automation testing.

Speakers
LD

Lalita Damaraju

Facebook Network Team, Facebook
AR

Ashik Ratnani

Facebook Network Team, Facebook


Friday March 15, 2019 9:30am - 9:55am PDT
210 DH

9:30am PDT

Ownership and Control of Firmware in Open Compute Project Devices
A firmware owner controls what firmware is allowed to run on a device or platform. Firmware ownership affects the overall security of OCP devices, which, in turn, affects the security of the compute infrastructure in which the devices are deployed. This session describes secure and efficient methods of establishing, representing, and transferring control of firmware in a device or platform. It provides detailed examples of ownership transfers throughout the lifecycle of a device. Finally, it relates these examples to the OCP tenets of efficiency, scalability, openness, and impact.

Speakers
avatar for Elaine Palmer

Elaine Palmer

Senior Technical Staff Member, IBM T.J. Watson Research Center and IBM Academy of Technology
I work on secure (verified) and trusted (measured) boot firmware. My background is in device ownership transfer, hardware security modules, smart cards, FIPS 140-2, and security evaluations.


Friday March 15, 2019 9:30am - 9:55am PDT
Marriott Salon III

9:38am PDT

Accelerated uCPE Powered by Modularity
Silicom modular uCPE solutions were presented by AT&T last year following the uCPE spec. For 2019 - Silicom shares its designs and roadmap. We are now moving to the next generation of uCPE and MEC, this will include full open architecture that leverages modularity to enable HW assist / acceleration options focusing on Edge Analytics / Smart Edge / AI / deep learning to facilitate Image processing, Network QoS management, Smart CDN, Edge media On Demand any many more technologies focused on Edge computing leveraging hardware accelerated infrastructure

Speakers
BB

Boris Beletsky

AVP Emerging Technologies, Silicom
EB

Elad Blatt

CSO, Silicom


Friday March 15, 2019 9:38am - 9:56am PDT
212 CD

9:45am PDT

OCP NIC 3.0 Power with Intel® Ethernet Products
Intel has been a key contributor to the OCP NIC 3.0 specification. In this session, we will discuss the updated OCP NIC 3.0 specifications with a focus on the power delivery and Intel’s product line of OCP NIC 3.0 adapters.

Speakers
avatar for Thomas Ng

Thomas Ng

Hardware Engineer, Intel
Come talk to me about OCP NIC 3.0!
avatar for Paul Kappler

Paul Kappler

Hardware Engineer, Intel


Friday March 15, 2019 9:45am - 9:57am PDT
210 A

9:45am PDT

SONiC Event Driven Dataplane Telemetry
We present a method for performing accurate and efficient telemetry based on dataplane events. Instead of capturing large volumes of statistical data (sflow or INT), we target specific network events such as packet drops, physical layer drop, buffer congestion and latency thresholds. An event record is generated with the truncated packet along with a cause or reason. An optional local aggregation mechanism is proposed that will reduce the network load by filtering any redundant data. For example, in a buffer tail drop, only a single representative packet per flow need be sent with relevant counters. A mechanism for integration in SONiC will be presented.

Speakers
MK

Matty Kadosh

Principle Architect, Mellanox Technologies
avatar for Aviad Raveh

Aviad Raveh

VP, BD technology, Mellanox Technologies
Aviad is the VP, BD technology, responsible of driving the Ethernet switch business development. Aviad collaborates with the biggest T1 and T2 cloud and web companies in the US and China. Until recently, he led the Ethernet switch software architecture and was responsible for building... Read More →


Friday March 15, 2019 9:45am - 9:57am PDT
Marriott Salon IV

10:00am PDT

Day 2 Welcome
Speakers

Friday March 15, 2019 10:00am - 10:10am PDT
Grand Ballroom

10:10am PDT

Enabling High Compute Density through Open19 Technology - Presented by LinkedIn
Speakers
avatar for Zaid Ali Kahn

Zaid Ali Kahn

Head of Infrastructure Engineering, LinkedIn
Zaid Ali Kahn is the head of Infrastructure Engineering at LinkedIn. His teams are responsible for architecture, engineering and operations for Networks, Data Centers, compute and storage serving over 500M members.Prior to LinkedIn, Zaid held several senior technical positions in... Read More →


Friday March 15, 2019 10:10am - 10:30am PDT
Grand Ballroom

10:31am PDT

Building Resiliency Through a Collaborative Ecosystem
As applications are migrating closer to the user and the data, data center architectures are becoming more complicated. To deliver these complex edge networks, the ecosystem needs to change the way in which it is working together.  The edge’s unique challenges can only be fully addressed by a collaborative environment that encompasses all the key partners – physical infrastructure vendors, system integrators, IT equipment manufacturers, managed service providers, and the customer.  Further, as data lakes with massive amounts of normalized data are created, we can take that collaborative environment to the next level by developing open APIs.  Implementing open APIs and allowing 3rd parties to access the data lakes enables the deployment of open-sourced applications and strengthens the edge ecosystem.  This presentation will highlight three keys to achieving a resilient edge:
  • A collaborative ecosystem that includes the customer is necessary
  • Cloud-based software is imperative to managing the edge
  • Resiliency is strengthened through open APIs for data lakes allowing open-sourced applications 

Speakers
avatar for Steven Carlini

Steven Carlini

Vice President, Innovation & Data Center, Schneider Electric
Steven Carlini is the Vice President of Innovation and Data Center for Schneider Electric. Steven is responsible developing integrated solutions and communicating the value proposition for Schneider Electric’s data center segment including enterprise and cloud data centers. A frequent... Read More →


Friday March 15, 2019 10:31am - 10:51am PDT
Grand Ballroom

10:51am PDT

Adopt OCP from AI with Agility and Speed - Presented by Inspur
For hyperscale data centers, OCP members are involved in helping to optimize racks to better handle the intensive, data-driven applications such as machine learning and artificial intelligence (AI).  As open community is committed to creating open hardware solutions working with data centers and their customers, competition is based on a vendor’s ability to respond to industry demand and emerging technologies that bring platforms to market more quickly and efficiently.  VP of Inspur Group, John Hu, will discuss how Inspur embraces openness and the key aspects in driving OCP based AI adoption, as well as how has Inspur’s JDM model deployed open sourced AI solutions with agility, speed and efficiency, to both cloud and enterprise customers.  John will also share Inspur’s new contributions to OCP in 2019.

Speakers
avatar for John Hu

John Hu

VP, Inspur Group, Inspur
Mr. John Hu currently serves as Vice President of Inspur Group Co., Ltd, Vice Director of State Key Laboratory of High-end Server & Storage Technology, Vice President of Inspur Electronic Information Industry Co., Ltd, and General Manager of Inspur Power Commercial Systems Co., Ltd... Read More →


Friday March 15, 2019 10:51am - 11:11am PDT
Grand Ballroom

11:12am PDT

OCP Adoption in Telco Ecosystem
The adoption of OCP hardware & software in the Broadband Access space by Service Providers in joint collaboration with ONF, System Integrators and the open-source community has picked up pace over the last year. As these solutions enter field-trails and production at Tier-1 network operators, AT&T would like to invite carriers around the globe to participate in this highly beneficial model of open-spec hardware plus open-source software development, curated by SPs, architecturally guided by ONF, and co-developed by the community. AT&T will also discuss how the work in the OCP Telco group and at ONF has accelerated deployment of new technology and enhanced competitiveness for their business. ONF will then highlight two specific solutions - Trellis and SEBA - built on OCP products for the broadband access market, that fall under ONF's CORD (Central Office Re-architected as a Datacenter) umbrella of projects. And finally, the Flex Cloud and Communication Solutions group will describe their contributions toward integration and automated validation of ONF solutions built on OCP platforms.

Speakers
avatar for Abdellah Tazi

Abdellah Tazi

Director Member of Technical Staff, AT&T
Abdellah Tazi is Director Member of Technical Staff for Converged Access and Device Technology at AT&T. Abdellah has Masters in Computer Information Systems Management. In his current role at AT&T, Abdellah has responsibility of corporate strategy for technology evolution through... Read More →
avatar for Saurav Das

Saurav Das

VP of Engineering, Open Networking Foundation (ONF)
Over 20 years in networking & communications, Saurav has worked on systems as diverse as network virtualization at Big Switch Networks, datacenter switches at Arista, and WDM optics at Enablence. He currently serves as VP of Engineering at the Open Networking Foundation, where he... Read More →
avatar for Christopher Winslow

Christopher Winslow

Senior Vice President & General Manager, Global Cloud & Data Center Business Unit, Flex
Chris Winslow is Senior Vice President and General Manager of the Global Cloud and Data Center business unit at Flex (NASDAQ: FLEX).  The Flex Global Cloud and Data Center business unit delivers innovative solutions, design, manufacturing, and global deployment models for the world’s leading providers in the cloud, enterprise data center, SaaS, and next generation internet, security, and st... Read More →


Friday March 15, 2019 11:12am - 11:37am PDT
Grand Ballroom

11:38am PDT

Huawei - Ushering OCP to New Frontiers
Speakers
avatar for Francis Lam

Francis Lam

Director, Intelligent Computing Product Management, Huawei Technologies
Francis brings 20+ years of IT industry experience specialized in server systems design.  Before joining Huawei Enterprise USA as Director of Product Management in 2015, Francis served in Huawei US R&D Center since 2011 as a senior server architect. Francis is responsible of driving... Read More →


Friday March 15, 2019 11:38am - 11:58am PDT
Grand Ballroom

11:58am PDT

Closing Remarks
Speakers
AH

Archna Haylock

Community Director, OCP


Friday March 15, 2019 11:58am - 12:00pm PDT
Grand Ballroom

12:00pm PDT

Lunch
Friday March 15, 2019 12:00pm - 1:00pm PDT
Lobby
  Meals
  • YouTube Link -

12:00pm PDT

Expo Hall Open
Friday March 15, 2019 12:00pm - 4:30pm PDT
Expo Hall 1
  Expo Hall
  • YouTube Link -

12:50pm PDT

Open Cloud, Open Access Solutions - Presented by Cambridge Industries
Introduction of Cloud and Access networking evolution, large scale JDM/ODM business model and open platforms/solutions, CIG’s contribution in Data Center Switch and Open OLT

Speakers
MX

Michael Xin

Cambridge Industries


Friday March 15, 2019 12:50pm - 1:05pm PDT
Expo Hall 1

1:00pm PDT

Immersed Computing® for OCP
This session is focused on what the Asperitas solutions are, how these are impacting the industry and the full range of products which address every aspect of immersion technology to facilitate a holistic end-to-end solution.

Speakers
avatar for Rolf Brink

Rolf Brink

CEO, Asperitas
Product development for cloud architectures and datacentre infrastructure, combined with a passion for cleantech innovations like liquid cooling, are the foundation of the developments which drive Asperitas.Asperitas is a clean-tech company focused on greening the datacentre industry... Read More →



Friday March 15, 2019 1:00pm - 1:25pm PDT
210 E

1:00pm PDT

RunBMC: A Hardware Interface for BMC Cards for OCP
We propose an openBMC development platform that would be the actual baseboard management controller platform and hardware you could leverage in any OCP design. At a high level it would consist of a BMC Mezzanine daughter card routing all major signals through a small pitch connector allowing a reusable, standardized footprint for OCP hardware designs. An additional development board would allow developers access to the BMC mezzanine early in the HW lifecycle, increasing openBMC contributions and adoption. The daughter card itself, which we call openBMC, would provide flexibility, reusability, and added velocity for server and switch designs in the Open Compute project.

Speakers
avatar for Jared Mednick

Jared Mednick

Hardware Engineer, Salesforce
avatar for Eric Shobe

Eric Shobe

Hardware Engineering Manager, Dropbox


Friday March 15, 2019 1:00pm - 1:25pm PDT
Marriott Salon I&II

1:00pm PDT

CIG CS6436-56P OCP Switch Proposal
Speakers
HZ

Harrison Zou

Product Architect, Cambridge Industries Group USA, Inc


Friday March 15, 2019 1:00pm - 1:25pm PDT
211 CD

1:00pm PDT

The Linux Kernel, Ecosystem and Community for Open Switch Hardware
The Linux kernel has been doing network hardware offloads and acceleration for decades (Nics, smart-nics, wireless AP’s and many other places). Because of this, the kernel has had the infrastructure and the right abstractions to recognize and register a networking hardware device since the early days. And this infrastructure has matured over time. Today Linux enablement is the best way to get faster adoption of your hardware. And also possibly the fastest way to go-to-market. In this talk we look at Open switch hardware's journey into the Linux kernel, ecosystem and community. We will show how embracing Linux for your open switch hardware can simplify and unify your networking infrastructure. Linux kernel has abstractions for supporting switch hardware. We will talk about the support for switch hardware in the Linux kernel and walk through many cases of collaborative development in the Linux community that made this possible. In the recent years the Linux kernel has also received many networking updates to make it a first class network operating system for a data center switch hardware. These include VRF, MPLS, VxLAN, E-VPN and many others. We will cover details on these updates. Today one can build a open E-VPN data center fabric with open switch hardware, Linux kernel and other open software from the Linux ecosystem. How cool is that ?. Running native Linux allows you to now also extend the same E-VPN fabric to servers or hypervisors. We will show how benefits of unifying your networking infrastructure with open hardware and Linux are endless: your provisioning and operational tools are developed and maintained for the same operating system and come from the same Linux ecosystem. We will conclude with a few details on the next big things coming in the Linux kernel and community to support the next generation switch hardware!

Speakers
RP

Roopa Prabhu

Director Engineering, Cumulus Networks


Friday March 15, 2019 1:00pm - 1:25pm PDT
210 DH

1:00pm PDT

OCP NIC 3.0 PCIe Electrical Compliance Test Fixture Update and Demo
Following to last year’s PCIe Electrical Compliance testing kick-off for OCP eco-system, please join us to learn more about how Dell-EMC, UNH-IOL, Keysight and Facebook have collaborated to enable standardized PCIe electrical compliance testing for the new OCP NIC 3.0 form factor. The session will provide insights into both the test fixtures and test methodology along with a live demo.

Speakers
avatar for Rick Eads

Rick Eads

Principal Engineer, Keysight
MK

Michael Klempa

Technical Manager, Ethernet and Storage, UNH-IOL
avatar for Jon Lewis

Jon Lewis

Distinguished Engineer, Dell EMC
RS

Richard Solomon

PCI-SIG Vice-President and Co-chair Compliance Workgroup, Synopsys
Richard Solomon is the Technical Marketing Manager for Synopsys' DesignWare PCI Express Controller IP.  He has been involved in the development of PCI chips dating back to pre-1.0 versions of the PCI spec. Richard architected and led the development of the PCI Express and PCI-X interface... Read More →


Friday March 15, 2019 1:00pm - 1:25pm PDT
210 A

1:00pm PDT

OCP 44OU Rack
The 44OU high capacity rack was designed to support a combination of network and server equipment. Two easily interchangeable configuration scenario's are supported offering flexibility for all customers.

Speakers
avatar for Steve Moore

Steve Moore

Project Engineer, Rittal
Contact Information: Steven Moore Project Engineer Phone: +44 01752 207 654 Email: smoore@rittal-csm.co.ukI have worked for Rittal over the past year and a quarter, starting out as a Design Engineer and transitioning to a Project Engineer. I am now the Project Engineer of one of Rittal’s... Read More →
avatar for Alex Lin

Alex Lin

Sr. Technical Product Manager, Penguin


Friday March 15, 2019 1:00pm - 1:25pm PDT
210 BF

1:00pm PDT

Convergence in Open Hardware Platforms
OCP Community is expanding its reach to consolidate different open hardware platforms (including Project Olympus, Open19 and OpenPower solutions) into the community to provide more selections and alternatives to datacenters of all sizes to scale globally. Inspur will share actual use cases from our customers, benchmarks for different application workloads and provide insights for optimization using different platforms. Inspur will also share new compute, storage, GPU server designs and OpenBMC hardware management contributions into the OCP community to drive adoption.

Speakers
avatar for Dolly Wu

Dolly Wu

Vice President/GM, USA Cloud Datacenter, Inspur Systems
Dolly Wu is the Vice President and GM of Inspur Systems, Inc. Dolly has over 20 years of experience in the high technology industry. Prior to joining Inspur, Dolly has been instrumental for driving business for leading technology companies, such as Hyve Solutions, Supermicro and... Read More →


Friday March 15, 2019 1:00pm - 1:25pm PDT
210 CG

1:00pm PDT

SONiC - Latest Update and Roadmap Forward
Since 2017 summit, a lot of new developments were added to SONiC, e.g. feature wise, operation maturity, new scenarios, etc. We are going to update the industry the last status of SONiC and share what is new coming on the SONiC roadmap.

Speakers
avatar for Xin Liu

Xin Liu

Principal Product Manager, Microsoft Azure, Microsoft
LY

Lihua Yuan

PARTNER SOFTWARE ENG MGR, Microsoft
Lihua has been with Microsoft for 10 years, working on building hyper scale network since 2008. He got his Ph.D in Computer Science from UC Davis. He has around 11 publications, and a rich experience in speaking on all sorts of conferences, from Sigcomm, to INFOCOM, OCP, ONS etc... Read More →


Friday March 15, 2019 1:00pm - 1:25pm PDT
Marriott Salon IV

1:00pm PDT

Case Study: 1OU Open Rack Server and Dense Rack Level Configurations
This study analyzes benefits and challenges in case if density of Open Rack would be increased.
Study compares possible server and rack level configurations if Open Rack triplet based server
height would be changed from 2OU to 1OU.
 
This study analyzes both single server level challenges related to configuration and cooling
limitations as well as rack level challenges related to power distribution, rack weight increase,
additional switch needs, cabling increase etc.

Speakers
ST

Samuli Toivola

R&D Squad Group Lead, Nokia


Friday March 15, 2019 1:00pm - 1:25pm PDT
212 CD

1:00pm PDT

Panel Discussion: 4 Principles to Realize Benefits of OCP
The facilities of today and the facilities of tomorrow must support a wide-range of use-cases. The existing published industry standards are not relevant to all of the use-cases. A panel of facilities focused leaders discusses the current standards and specifications they utilize and the vision for the future.

Speakers
RB

Robert Bunger

Program Director in CTO Office, Schneider Electric
avatar for Mike Moore

Mike Moore

Edge Cloud Platform Product Sales, OpenEdge Sub-Project Lead, Nokia
Over 20 years of R&D experience in both the Wireless and Defense Industries prior to transitioning into a customer facing sales and support role.  Now the Western Hemisphere Regional Product Manager for Nokia’s AirFrame Data Center Portfolio, with a focus on creating highly available... Read More →
RL

Russ Lindsay

SVP, Salesforce
avatar for Stijn de Kruijf

Stijn de Kruijf

Data center facility design expert, Royal HaskoningDHV
avatar for Dale Sartor

Dale Sartor

Staff Engineer, Building and Industrial Applications, Lawrence Berkeley National Laboratory
Dale Sartor, P.E., is a staff scientist/engineer at the Lawrence Berkeley National Laboratory (LBNL) working on building energy efficiency applications. Mr. Sartor has an AB in Architecture, and an MBA. He is a mechanical engineer with over forty years of professional experience in... Read More →
avatar for Mike Edie

Mike Edie

Mechanical Engineer, Strategic Engineering and Design, Facebook


Friday March 15, 2019 1:00pm - 1:50pm PDT
211 AB

1:10pm PDT

NVMe Optimized Boot Solution - Presented by Lite-On Storage
LITE-ON proposes to integrate popular M.2 NVMe drives and NVMe RAID controller into a boot solution package. This boot solution package is optimized for redundancy, compact form factor, and flexibility. These features are important for the next generation server platforms that promote increased density without sacrificing high availability.

Speakers
JC

Jeffery Chang

Principal Engineer, Lite-On Storage


Friday March 15, 2019 1:10pm - 1:25pm PDT
Expo Hall 1

1:20pm PDT

SMR and ZNS – Two Sides of the Same Coin
How to efficiently use large storage capacity devices in the data center with an eye to reducing costs ($/TB).

Speakers
avatar for Lee Prewitt

Lee Prewitt

Principal Hardware Program Manager, Microsoft
Lee has over 25 of storage industry experience having worked on many types of storage devices ranging from Magneto-Optical to CD/DVD/Blu-ray to spinning rust to Flash. Lee started at Microsoft in 2008 working in the Windows and Devices Group where he was responsible for many of... Read More →


Friday March 15, 2019 1:20pm - 1:38pm PDT
212 AB

1:30pm PDT

Scaled Troubleshoot Process for Optical Interconnects at Facebook Data Centers Using Network Switches: BER Illusion Methodology (BIM)
To support the increased number of users (over 2.27 billion monthly active users) and data usage, Facebook has been deploying 100G CWDM4 technology (QSFP 28). Currently, when there is a link failure, it is a very tedious process to troubleshoot the optical interconnects with the existing tools. These troubleshoot tools are very expensive and can characterize one optical transceiver at a time. This is a very time consuming process and the operational cost is very high. To reduce the operational costs and to optimize the troubleshoot time, we have developed a unique tool called “BER Illusion Methodology” (BIM) which is based on OCP network switches. This methodology can be used to troubleshoot 128 optical interconnects simultaneously in less than 5 minutes including all necessary information such as Transmitter power, Transmitter bias current, Receiver power, Case Temperature, complete BER test, supplier details, etc. The number of troubleshoot ports can be scaled, just by adding more switches if needed in the topology. This endeavor describes a post production troubleshoot/sustaining process which is being implemented to identify and characterize optical interconnect failures at our data centers so that we can reduce the “No Trouble Found (NTF)” numbers significantly and with minimal effort.

Speakers
AC

Abhijit Chakravarty

Manufacturing and Quality Engineer, Facebook
EG

Edward Galley

Director, Manufacturing Quality Engineering, Facebook
AP

Ashwin Poojary

Systems Engineer, Network, Facebook


Friday March 15, 2019 1:30pm - 1:42pm PDT
211 CD

1:30pm PDT

Flex Helsinki, 2OU OpenRACK Built for High-density, High-reliability Telco Cloud
Join us to learn about Helsinki, Flex’s latest storage platform to gain OCP Inspired™ status. Interact directly with the engineering team responsible for designing the platform. Helsinki is a 2OU rack-mount 45-drive JBOD built for high-density, high-reliability Telco cloud applications. It is NEBS compliant and meets the most stringent safety and EMC standards globally, along with shock and vibration, thermal, fire resistance, and seismic requirements. Helsinki supports both 3.5” and 2.5” drives, SAS/SATA HDDs and SSDs

Speakers
SG

Srinivas Gadgil

Chief Engineer – HW & SW platforms, Flex


Friday March 15, 2019 1:30pm - 1:44pm PDT
212 CD
  EW: Telco & openEDGE
  • YouTube Link -

1:30pm PDT

The 8 Best-Fit Workloads for Enterprise QLC SSDs - Presented by Micron
With breakthrough storage technologies like Quad-Level Cell (QLC) SSDs now readily available, storage developers need to become more sophisticated in their endurance needs. Based on real results and countless customer uses for the world’s first QLC SSD, Micron will show you the 8 best ways to deploy QLC SSDs and how to save and adopt them in your systems.

Speakers
SH

Steve Hanna

Sr Product Marketing Manager, Micron


Friday March 15, 2019 1:30pm - 1:45pm PDT
Expo Hall 1

1:30pm PDT

OCP-ACS Door Heat Exchanger Concepts/Specification Update
Door heat exchanger (HEx) is one of the new work streams under Rack and Power ACS (advanced cooling solutions) sub-group. This presentation is meant to give a high-level overview of ACS Door HEx concepts and their potential applications in server racks and data centers, as well as providing updates to the community on the specification for the same.

Speakers
avatar for John E. Fernandes

John E. Fernandes

Thermal Engineer, Facebook


Friday March 15, 2019 1:30pm - 1:55pm PDT
210 E

1:30pm PDT

Performance Oriented Firmware Implementation for Error Handling in Large Scale Data Center
Hardware failures always happen in large scale data center. It is critical how to minimize the performance impact due to hardware failures. Facebook has implemented performance oriented firmware on system level to address this need in error handling

Speakers
CD

Chris Davis

Core Data Production Engineer, Facebook
ZY

Zhengyu Yang

Hardware System Engineer, Facebook


Friday March 15, 2019 1:30pm - 1:55pm PDT
Marriott Salon I&II

1:30pm PDT

Open Networking (white box) in the Enterprise
White box switches and open networking are no longer only for the hyper scale companies with giant IT organizations. White box in the enterprise is possible, given the right mindset, and a well thought out plan (after much lab testing)... This session discusses the pro's and con's, challenges and pitfalls, and eventual success of white box switch deployments at a medium size enterprise.

Speakers
MT

Matt Turner

Data Center Network Manager, Qualcomm Inc.


Friday March 15, 2019 1:30pm - 1:55pm PDT
210 DH

1:30pm PDT

OCP NIC Thermal Test Fixture Standardization and Demo (Server WG)
As OCP NIC form factor evolving into 3.0 version, there are rising needs to standardize thermal characteristic testing methodology among NIC suppliers, system suppliers and hyper-scale users. This session will be a joint session between Dell EMC and FB to talk about the design of the thermal test fixture and how to leverage it for further development of NIC 3.0 spec with a live Demo.

Speakers
avatar for Yueming Li

Yueming Li

Thermal Engineer, Facebook
avatar for Hemal Shah

Hemal Shah

Distinguished Engineer, Broadcom
Hemal Shah is a Distinguished Engineer and Systems/Software/Standards architect in the Compute and Connectivity (CCX) division at Broadcom Inc. He leads a team of architects. In his role, he is responsible for the product architecture definition of Ethernet NIC products, NIC integration... Read More →
avatar for Jon Lewis

Jon Lewis

Distinguished Engineer, Dell EMC


Friday March 15, 2019 1:30pm - 1:55pm PDT
210 A

1:30pm PDT

Storage Security from A to Z
Data security is becoming an increasingly lucrative target for cyber criminals. Worldwide revenue for data theft is estimated to be approaching $1 trillion per year. Storage vendors including Seagate have invested heavily in data security from secure supply chain, to securing data at rest to end of life solutions like instant secure erase. These are all based on industry and government standards and best practices.

Collaboration with OCP Security efforts by device makers is essential to a full defense-in-depth data security strategy that defends against the possibility of malware or defects from privileged device related firmware and software.


Speakers
avatar for Arie van der Hoeven

Arie van der Hoeven

Principal Product Manager - Cloud Ecosystem Lead, Seagate


Friday March 15, 2019 1:30pm - 1:55pm PDT
Marriott Salon III

1:30pm PDT

Open19 Platform Deep Dive
Design overview and deep dive into Open19 architecture, ecosystem, and lessons learned from production deployments.

Speakers
SZ

Sergiy Zhuk

Director, Hardware Engineering, LinkedIn
BP

Brad Peterson, Data Center Engineering

Sr. Staff Data Center Engineer, LinkedIn


Friday March 15, 2019 1:30pm - 1:55pm PDT
210 BF
  EW: Rack & Power
  • YouTube Link -

1:30pm PDT

How Effective was OCP Adoption in Yahoo! Japan Infrastructure as Non-Board Member of OCP
Yahoo! JAPAN is the Top-Level Japan's Contents Service Provider. And we have adopted OCP solutions at our infrastructure in Japan's DC since 2016. This session provide "Actually, How effective by adopting OCP as Non-Board Member of OCP". Especially, I'd like you to refer to session for People who are considering OCP which includes like Non-Board Member of OCP, as same infrastructure scale as Yahoo! Japan. This session will include below topics - CAPEX/OPEX figures from Yahoo! Japan Case - In Japan, Datacenter standard for OCP / Power standard for OCP - In our datacenter, how to delivery and install of OCP - operation cost comparison between EIA19 and OCP from actual data I'd like to emphasize that OCP is one of the effective solution for Non-Board Member and U.S. Datacenter.

Speakers
avatar for Kazuhide Fujimi

Kazuhide Fujimi

Server Infrastructure Architect, Yahoo Japan Corporation
Kazuhide Fujimi is a Server Infrastructure Architect at Yahoo! Japan and has responsibility for procurement, testing, QA and troubleshooting of server hardware. Yahoo! Japan is one of the famous CSP in Japan and has several data centers in Japan as on-premise Infrastructure envir... Read More →


Friday March 15, 2019 1:30pm - 1:55pm PDT
210 CG
  EW: Server
  • YouTube Link -

1:30pm PDT

Alibaba's SONiC Development for Large Scale Deployment and Operation
Alibaba has been an active participant of SONiC community since its early stage. Over the past two years, we have developed a rich set of features and tools around the SONiC system to make it easy to deploy and operate in our large scale data centers. In this session, we will share our development efforts, and our experiences for SONiC deployment and operations. which will cover core system optimizations, monitoring, management and other operational tooling around SONiC system.

Speakers
GW

Guohui Wang

Engineering Director, Alibaba


Friday March 15, 2019 1:30pm - 1:55pm PDT
Marriott Salon IV

1:40pm PDT

Toward the Large Deployment of Open Channel SSD
Alibaba is working on development and deployment of Alibaba Open Channel (AOC) SSDs to address the challenges on the storage devices in today's hyper-scale data centers. Increasing proliferation of Artificial Intelligence, E-commerce, Internet-of-Things, Big Data and Cloud applications is leading to highly diversified workloads and usage cases. These applications can have drastically different I/O patterns, and performance/Quality-of-Service (QoS) targets. Traditional storage architecture is designed around standardized hardware that provides generic block I/O interface, thus has the advantages of portability and backward compatibility. However, it is very challenging for the traditional storage architecture to meet the specific requirements with various usage models. To address this gap, Alibaba is focused on the development and deployment of open channel SSDs since from 2017. Open channel provides the capability of software/hardware co-optimization and storage stack customization based on the applications. In this work, we present our progress towards the large scale deployment of open channel SSDs in hyper-scale data centers. We have defined the Alibaba Open channel (AOC) SSD spec, and built the open channel SSD platform with our industry partners. On top of this platform, we develop the multi-stream capability to achieve reduced write amplification and boosted performance with a simplified Flash-Translation-Layer implementation owing to the co-optimization with the application. We also develop the multiple physical partitions feature with open channel SSDs to achieve the very low read latency in mixed read/write workload through the separation of read and write in search applications.

Speakers
FZ

Feng Zhu

Staff Engineer of Storage, Alibaba Group


Friday March 15, 2019 1:40pm - 1:58pm PDT
212 AB

1:45pm PDT

An Introduction to the Innovative and Portable iQSFP Tester
Speakers
AC

Abhijit Chakravarty

Manufacturing and Quality Engineer, Facebook
avatar for Jimmy  Leung

Jimmy Leung

Mechanical Engineering Manager, Facebook, Inc.
VZ

Vincent Zeng

MQE, Facebook


Friday March 15, 2019 1:45pm - 1:58pm PDT
211 CD

1:45pm PDT

How ONF Builds Solutions on OCP Products for Telco/Cableco Broadband Access
Over the last 3 years, ONF has been building open-source solutions on top of OCP accepted hardware and software for the telco/cableco broadband access market. As these solutions are now in field-trails and production at Tier-1 network operators, ONF would like to introduce these solutions to the larger OCP community. This talk covers two specific ONF solutions - Trellis and SEBA. Trellis is a multi-purpose leaf-spine fabric designed for NFV & access/edge applications. It is SDN-based, built with bare-metal switches and open-source software. Trellis builds an extensive feature set on top of OCP accepted hardware from EdgeCore with several ASICs from Broadcom including Trident, Tomahawk and Qumrans, as well as OCP accepted software like ONIE and ONL. Trellis is currently deployed in production. SEBA, or SDN Enabled Broadband Access, is a self-contained pod solution for disaggregated Passive Optical Networks (PONs). It was created to provide an architecture pattern and reference implementation for developing solutions for carrier broadband access. In particular, the reference implementation is built on an OCP approved specification for AT&T's 16 port XGS-PON vOLT with hardware from EdgeCore. SEBA is currently deployed in field-trials. Both SEBA and Trellis are completely open-source and fall under ONF's CORD (Central Office Re-architected as a Datacenter) umbrella of projects.

Speakers
avatar for Saurav Das

Saurav Das

VP of Engineering, Open Networking Foundation (ONF)
Over 20 years in networking & communications, Saurav has worked on systems as diverse as network virtualization at Big Switch Networks, datacenter switches at Arista, and WDM optics at Enablence. He currently serves as VP of Engineering at the Open Networking Foundation, where he... Read More →


Friday March 15, 2019 1:45pm - 1:59pm PDT
212 CD

1:50pm PDT

Juniper Open Initiatives Overview - Presented by Juniper Networks
Juniper has been an active participant in various open source communities for several years. This session provides an overview into the various initiatives that Juniper has engaged in, and has on display at this OCP Global Summit. Join Juniper Fellow Raghavendra Mallya as he covers: •SONiC support for both merchant and Juniper ASICs for both single chip and multi-chip systems •JUNOS control plane (cRPD) running as a software micro-service, providing management, routing, programmability and rich telemetry in a container •Active initiatives developing open APIs at every level of the stack, spanning from OpenR at the application layer to P4 at the forwarding layer

Speakers
RM

Raghavendra Mallya

VP Juniper Fellow, Juniper Networks


Friday March 15, 2019 1:50pm - 2:05pm PDT
Expo Hall 1

2:00pm PDT

Telco & openEDGE Validation of CORD SEBA Reference Design on the OCP Platform
Since the first release of CORD in 2016, the Flex Cloud and Communication Solutions group has collaborated with the ONF QA group in the integration and automated validation of CORD on OCP platforms. A new reference design known as SEBA (SDN enabled broadband access) will launch in December 2018 as part of CORD 6.0. SEBA is the residential use case and provides optimizations so that network traffic can run 'fastpath' straight to the backbone without requiring VNF processing on a server. The SEBA infrastructure setup at the Flex CCS CloudLab consists of a full CORD POD deployed via Jenkins on OCP platforms, a fabric switch, an interconnected OLT, a splitter, 4 ONUs, and 16 subscribers to validate the SEBA traffic flow. In this talk, we will demonstrate how we validated the SEBA reference design on OCP. This will include a presentation of the functional, stability, scalability, and soak tests executed on the OCP architecture and a discussion of the results.

Speakers
avatar for Siddharth Gogar

Siddharth Gogar

SDN Developer, Flex
• Working on deployments of Telco projects including CORD, OPNFV, ONAP and private cloud deployments of RedHat OSP, Fuel and Nephoscale. Integrate VNF onboarding solutions and industry standard benchmarks used for telco workloads such as yardstick and spirent tools with private... Read More →


Friday March 15, 2019 2:00pm - 2:14pm PDT
212 CD

2:00pm PDT

From Open-Channel SSDs to Zoned Namespaces
Open-Channel Solid State Drive architectures are adopted rapidly by hyper-scales, all-flash array vendors, and large storage system vendors. The versatile storage interface admits solid state drive to expose essential knobs to control latency, I/O predictability, and I/O isolation. The rapid adoption has created a diverse set of different Open-Channel SSD drive specs, that each solves the need of a single or few users. However, the specifications are yet to be standardized. The Zoned Namespaces (ZNS) Technical Proposal in the NVMe workgroup is developing an industry standardization for these types of interface. Creating a foundation of which we can build a robust software eco-system on top and streamline implementation efforts. This talk covers the motivation, characteristics of Zoned Namespaces, possible software improvements, and early results to show off the effectiveness of these types of drives.

Speakers
MB

Matias Bjørling

Director, Solid-State System Software, Western Digital


Friday March 15, 2019 2:00pm - 2:18pm PDT
212 AB
  EW: Storage
  • YouTube Link -

2:00pm PDT

Door HEX, LCP Hybrid and Connection on the OCP V2 Rack
Presentation of the adaption from the HEX to OCP rack. ACS, HEX working group

Speakers
avatar for Resul Altinkilic

Resul Altinkilic

Project Manager Global Key Accounts IT, Rittal
Project Management Global Key Accounts IT, working on High Performance Computing / Cooling and also on Edge Solutions. Involved in the OCP Community since 2017 (Open Compute Project) and the development of OCP Rear Door Solutions.Former Project Manager for the Automotive and automatization... Read More →


Friday March 15, 2019 2:00pm - 2:25pm PDT
210 E

2:00pm PDT

ISO Containers are NOT a Good Data Center Solution
When exploring modular data center solutions, the ISO shipping container is frequently part of the discussion. Modular, purpose built enclosures are a great solution for data centers. Modular hybrid constructed buildings like CyrusOne and Microsoft, enterprise data systems, and telecom systems are all a perfect fit for modular, but not shipping containers.When building a modular data center, the walls are not the bulk of the cost, and should not be a limiting factor. It is true that containers are cheap and readily available, 80% of them are made in China to ship goods all over the world, and the Americas get a steady surplus of these metal boxes. To make them suitable for a data center (even a limited one with few racks) requires significant modification. Cleaning, repair, painting, insulating, door installation and more must be done to provide acceptable white space. These modifications are time consuming and costly, this quickly eats up the initial savings from that "cheap" shipping container. Purpose built modular data centers are the safest, most efficient, and most functional way to add capacity for any network. In this session, participants will learn the details of modular enclosures and limiting factors of ISO containers by looking at efficiencies, code requirements, and transportation costs.

Speakers
avatar for Rob Coyle

Rob Coyle

Director of Sales and Marketing, PCX Corporation
As the son, grandson, and brother of carpenters, my education in construction began early. This background, combined with my passion for innovative technology solutions, led me to modular construction. The construction industry has been historically slow-to-change, yet the problems... Read More →


Friday March 15, 2019 2:00pm - 2:25pm PDT
211 AB

2:00pm PDT

How OCP Helped the Department of Energy Labs Place Nine Supercomputers in the Top500
As research becomes more complicated and expensive, the value proposition of open technologies is becoming more alluring. However, there is still a belief that the performance capabilities of open technologies cannot match the needs of leading-edge energy researchers, with their complex simulation and analysis needs. This presentation describes how open source, specifically Linux and Open Compute Project (OCP) technologies can be carefully – and cost-effectively – designed to meet even the rigorous goals of the U.S. Department of Energy’s (DOE’s) National Nuclear Security Administration (NNSA) as it helps secure the nuclear stockpile in accord with the Comprehensive Nuclear-Test-Ban Treaty (CTBT) between the U.S. and the former Soviet Union using simulation-based confidence (versus test-based confidence). The performance of these commodity-based supercomputers was so high that nine were ranked in the November 2018 Top500 list of most powerful computers on Earth. Yet the technology used was a turn-key OCP solution that is available to non-government clients, underscoring the performance to value that OCP solutions can provide other organizations.

Speakers
SM

Sidney Mair

Senior Vice President of Federal Systems, Penguin Computing


Friday March 15, 2019 2:00pm - 2:25pm PDT
210 A

2:00pm PDT

Manageability Security
A presentation on how to approach security for Management Hardware and Firmware. Understanding the threats to a system and how to approach defending against attacks from adversaries. This presentation will cover some of the challenges the industry is faced with today regarding manageability. Where hardware vendors are being asked to provide the maximum level of manageability while providing the highest level of security. This is especially important with the increasing threat landscape of computers being deployed in every aspect of our lives. This presentation will highlight some of the things to consider when designing hardware and firmware with manageability in mind.

Speakers
avatar for James Mihm

James Mihm

BMC FW Team Lead, Intel Corporation
James works in the Data Center Group on the architecture and design of Baseboard Management Controller Firmware with a focus on BMC Security.


Friday March 15, 2019 2:00pm - 2:25pm PDT
Marriott Salon I&II

2:00pm PDT

Real Examples of Telecom Services Built Using White Boxes
For decades, telecom services have been enabled almost exclusively by proprietary, vendor-supplied hardware and software, often bundled, resulting in little choice, customization, or flexibility for operators and at high cost. The disaggregation of hardware and software and the open sourcing of both, pioneered by OCP, has inspired the telecom industry to break out of its chains and adopt white boxes for servers, switches, various appliances, and the new Universal CPE as platforms for VNFs that eliminate the service specificity of the hardware. And as more telecom services, orchestration, and management are conducted in data-center environments, thanks to SDN and NFV, the OCP model becomes the one to emulate. In this talk we will describe some examples of working proofs of concept shown recently by MEF members and partners of white boxes and open-source or proprietary software to enable new, agile services and capabilities.

Speakers
avatar for Dan Pitt

Dan Pitt

Sr. VP, MEF Forum, and President, Palo Alto Innovation Advisors
Dan Pitt has been a leading voice of the SDNFV movement since its inception. He served as executive director of ONF from its public launch in March 2011 until September 2016 and since January 2017 has been senior vice president of MEF, driving technical and ecosystem development... Read More →


Friday March 15, 2019 2:00pm - 2:25pm PDT
211 CD

2:00pm PDT

Software Defined Fabric for OCP-based Leaf & Spine Switches
Similar to the goals outlined and achieved by Facebook at its Prineville, Oregon, data center facility to prove the value of the Open Compute Project, Kaloom was founded with the vision of providing the most automated and programmable data center networking fabric that will disrupt how data centers are built and managed by cloud providers, telcos and enterprises. The company has created a Software Defined Fabric (SDF) that leverages open networking OCP-compliant white boxes from vendors including Accton and Delta Networks, among others, to reduce CAPEX. Automation features reduce OPEX and minimize time to deployment and configuration errors while the programmable capabilities of its fabric future-proofs network investments compared to slower, more costly merchant silicon product cycles. The fabric leverages OCP’s Open Network Install Environment (ONIE) so that the installer can easily load the network OS across a variety of OCP-compliant switches, delivering greater choice and enabling operators to avoid vendor lock-in. Integrating vRouter, vSwitch, vGateway and other networking functions into a more streamlined data plane pathway enables data center operators to realize OCP’s vision for highly efficient, low-cost operations with dramatically increased performance. By utilizing a container-based architecture and collapsing, or offloading, data plane functions from numerous VMs, a Software Defined Fabric architecture can deliver an increase of up to 2x in throughput with a 7x reduction in latency, improving overall networking efficiency by a factor of 5-10x. This speaker will examine new type of SDN-agnostic networking fabric controller for the OCP switch topology management of spine, leaf and controllers; including virtual fabric creation with resource allocation and modification to deliver virtual network fabric slicing; and utilizing external interfaces. This presentation will explore the benefits of OCP-compliant open source software for networking switch gear leveraging ONIE, and highlight collaboration opportunities with other network focused software projects including OpenStack, OpenShift, Kubernetes and Linux.

Speakers
avatar for Thomas Eklund

Thomas Eklund

VP of Marketing and Strategy, Kaloom, Inc.
VP of Marketing and Strategy at Kaloom, Inc.Tom has a successful track record as an entrepreneur and as an executive, focusing on data driven businesses and networking. He has extensive experience operating global businesses, leading diverse cross functional teams and recruiting extraordinary... Read More →


Friday March 15, 2019 2:00pm - 2:25pm PDT
210 DH

2:00pm PDT

Go Forth and Modify: Fiano
In this talk we present fiano, Go-based tools that we created at Google and Facebook for manipulating UEFI images. Fiano is fast, scriptable, easy to use, and most importantly, does not require you to have UEFI source -- you can modify UEFI ROMs, removing most DXEs and, if desired, replacing them with a Linux kernel.

Speakers
avatar for Gan Shun Lim

Gan Shun Lim

Software Engineer, Google
RO

Ryan O'Leary

Software Engineer, Google


Friday March 15, 2019 2:00pm - 2:25pm PDT
Marriott Salon III

2:00pm PDT

Battery Backup Unit - Facebook Roadmap Exploration
This presentation is meant to give insight into the direction for Facebook's battery backup unit (BBU) design to be used in Open Rack. The material will cover the learning from the ORv2 Power supply and BBU design as well as the new features being enabled with the next generation product. In addition, Facebook's approach on some applications and functions such Power Loss Siren and peak power shaving, leveraging BBUs will be discussed.

Speakers
ZE

Zoh Ebrahimzadeh

Hardware Engineer, Facebook


Friday March 15, 2019 2:00pm - 2:25pm PDT
210 BF

2:00pm PDT

OCP Accelerator Module (OAM) System: An Open Accelerator Infrastructure Project
The OCP Accelerator Module (OAM) specification defines the form factor and interconnects for an open-hardware compute accelerator module.  A compliant Baseboard (UBB) along with a Tray, Chassis, and System Security and Control Module (SCM) form an open infrastructure for interoperable OAMs.

Speakers
avatar for Richard Ding

Richard Ding

Principal AI System Architect, Baidu
Richard Ding is Principal AI System Architect from Baidu Infrastructure Team. He leads architecture design of Baidu AI Computing Platform KongMing including AI computer X-MAN, high-performance parallel file system FAST-F, large scale and distributed communication engine Halo, as well... Read More →
avatar for Siamak Tavallaei

Siamak Tavallaei

Principal Architect, Microsoft Azure; CXL Consortium™ Technical Task Force Co-Chair, Microsoft
Siamak Tavallei is a principal architect and technologist instrumental in development and evolution of innovative multi-node, and multi-processor servers as well as technology initiatives in areas of storage/memory hierarchy and heterogeneous, distributed computing. He is a customer-driven... Read More →
avatar for Whitney Zhao

Whitney Zhao

Engineer, Facebook
Whitney is a seasoned hardware engineer leading AI/ML system design in Facebook. Whitney has led multiple hardware generations ranging from general purpose 2S system such as Tioga Pass to ML JBOG Big Basin systems, all of which have been contributed to OCP. She has been driving multiple... Read More →


Friday March 15, 2019 2:00pm - 2:25pm PDT
210 CG

2:00pm PDT

LinkedIn Adoption of OCP SONiC
Linkedin contribution and experience with OCP SONiC. Lessons learnt, development/maintenance case study, future looking features.

Speakers
ZX

Zhenggen Xu

SW Enginneer, Linkedin


Friday March 15, 2019 2:00pm - 2:25pm PDT
Marriott Salon IV

2:10pm PDT

Flex Discusses Emerging Technologies for the Next Generation Data Center - Presented by Flex
Flex discusses their view of new data center platforms with an eye toward disruptive emerging technologies.    

Speakers
avatar for Adam Tarnowski

Adam Tarnowski

VP Business Development for Cloud and Datacenter, Flex


Friday March 15, 2019 2:10pm - 2:25pm PDT
Expo Hall 1

2:15pm PDT

Open Edge Deployments
Edge datacenter facilities are often limited in terms of space, power and cooling. Due to the limitations, requirements of an edge datacenter solution are different from classical datacenters. In this presentation we will share information about edge datacenter solutions including open edge server specifications and design, requirements for open edge indoor rack solution, and open edge deployment scenarios.

Speakers
avatar for Mika Hatanpaa

Mika Hatanpaa

Head of R&D, Nokia
Mika is an experienced and passionate Head Of R&D with strong knowledge and experience in Telecommunications, Data Center Infrastructure, Cloud Infrastructure, Cloud Computing, Edge Cloud and Radio Access Networks, including 5G. Mika has proven track record in leading world class... Read More →


Friday March 15, 2019 2:15pm - 2:40pm PDT
212 CD

2:20pm PDT

Ethernet as the Next Storage-Fabric of Choice
PCIe Gen4/5 has a lot of performance promise but at the same time many scalability concerns. Ethernet, combined with NVMe-oF work, is poised to replace PCIe as the fabric of choice in large scale storage array and storage racks. In this presentation we will explore why Ethernet is finally ready to become the fabric-storage of choice for next gen large scale deployments.

Speakers
SN

Shahar Noy

Sr Director Product Marketing, Marvell
Shahar Noy is a Senior Director of product marketing for Data Center Storage Solutions at Marvell. Shahar joined Marvell’s superhero family in 2017 and other than occasionally saving the world he is busy resolving some of the biggest storage challenges in modern data centers. With... Read More →


Friday March 15, 2019 2:20pm - 2:38pm PDT
212 AB

2:30pm PDT

The Cloud Reloaded: How to run the SAME WORKLOAD with up to 60% LESS Hardware - Presented by Toshiba
The first generation of private cloud implementations were, by and large, a half-step. While new applications are now mostly built on top of scale-out databases, and containerized microservices are becoming the norm for stateless applications, outside of the large public CSPs, true dynamically orchestrated, self-service computing remains rare. Many talk a good game, but don’t really deliver.

What's holding up the show, and how do we get it moving again?

This talk will focus on the main technical hurdles to be overcome, and the key events and technologies that will usher in the cloud 2.0 era. And, we’ll detail the sources of some jaw-dropping financial rewards that await those who “go the distance” to running a true cloud infrastructure. Along the way we’ll provide fearless prognostications about some popular technologies that may have outlived their usefulness.

Speakers
JD

Joel Dedrick

VP & GM KumoScale - Shared, Accelerated Storage, Toshiba Memory America, Inc.


Friday March 15, 2019 2:30pm - 2:45pm PDT
Expo Hall 1

2:30pm PDT

Service Oriented 2-Phase Immersion Cooling with 48V Power Solution
To meet the upcoming high power density demand with high energy saving, Wiwynn developed an immersion cooling tank solution with 48V DC power delivery. One hundred OCP Tioga Pass server boards are immersed in a tank with 1.2m x 1.2m footprint. Two-phase immersion cooling technology is employed due to the high power density capacity and low Power Usage Efficiency (PUE) with warm water cooling. For easy service and high density design, each two boards are face-to-face compacted in a cartridge with one-hand released handle which is installed in the tank vertically to connect 48 V power bus bar. Sloped top cover design with 1.2 meter tank height design allows the front side service for the cartridge on rear side of the tank. Proprietary condenser system and water flow rate control are integrated in the tank management system to minimize vapor leakage during maintenance.

Speakers
avatar for Lentis Pai

Lentis Pai

Director, Wiwynn Corporation


Friday March 15, 2019 2:30pm - 2:55pm PDT
210 E

2:30pm PDT

Data Center Cooling Based on Predicting Power
Most datacenters today adjust the supply of cooling based on the awareness of cooling demand changes in the datacenter. Given that there is latency associated with meeting the new cooling demand generated, compute nodes or key components will be in overheating or overcooling status for a certain period of time. This not only impacts the power usage effectiveness(PUE), but also impacts performance due to the overheating induced throttling. This presentation provides a prediction approach of server power in the future based on monitoring power trends in prior periods. Applying this approach results in an intelligent cooling control system that can predicts the cooling demand in advance, then provides the proper cooling supply before the demand really happens thereby resolving the latency or reaction time issue. Utilizing this approach, enables both an improvement in the power usage effectiveness (PUE)by minimizing overcooling and also mitigates the risk of server overheating

Speakers
NA

Nishi Ahuja

Senior Principal Engineer, Intel
avatar for Mohan J Kumar

Mohan J Kumar

Fellow, Intel
Mohan Kumar is an Intel Fellow in the Data Platforms Group and the director of Cloud Architecture and Solutions in the Cloud & Enterprise Solutions Group at Intel Corporation. His organization is responsible for driving the architecture requirements for Intel Cloud customers.Kumar... Read More →
ES

Edmund Song

Software Architect, Intel Corporation


Friday March 15, 2019 2:30pm - 2:55pm PDT
211 AB

2:30pm PDT

Inspur Optimized Architecture for AI in OCP
This will be an in-depth discussion of Inspur's OCP AI product and application/use case: - an Ultra High Density GPU (4OU 16*GPU) for training or 16*FPGA for inference - Optimized architecture for AI to speed up the training process and flexible architecture to support multi-node based on application

Speakers
avatar for Han Wang

Han Wang

Principal Architect, Inspur
Dr. Wang is Principal Architect and Senior Client Product Manager at Inspur US. He has been in server, storage, cloud infrastructure industry for more than 16 years and has full experience and solid cross-field expertise in foundational research, engineering development, engineering... Read More →
GW

Gavin Wang

System Architect - Manager, Inspur


Friday March 15, 2019 2:30pm - 2:55pm PDT
210 A

2:30pm PDT

PMCI Standards for Hardware Management
The OCP Hardware Management Project creates designs and specifications to enable interoperable manageability for Open Compute platforms. Platform Management Components Intercommunications (PMCI) is a Working Group (WG) within DMTF that has developed standards to address “inside the box” communication and functional interfaces between the components of the platform management subsystem. PMCI standards can be leveraged for achieving interoperability between hardware management components of Open Compute Platforms. This talk will cover PMCI scope, work areas, published specifications, including the following important technologies: Management Component Transport Protocol (MCTP), Platform Level Data Model (PLDM), and Network Controller Sideband Interface (NC-SI). MCTP is a physical medium independent transport protocol designed to support communications between different platform management subsystem components. PLDM is a data model targeted for efficient communications among platform management subsystem components for accessing low-level platform inventory, monitoring, and control functions including SMBIOS data transfer, Redfish Device Enablement, BIOS control/configuration, FRU data transfer, and firmware update. NC-SI is a sideband communication protocol to enable the exchange of management data between the Management Controller and Network Controller.

Speakers
avatar for Patrick Caporale

Patrick Caporale

Distinguished Engineer, Lenovo
avatar for Hemal Shah

Hemal Shah

Distinguished Engineer, Broadcom
Hemal Shah is a Distinguished Engineer and Systems/Software/Standards architect in the Compute and Connectivity (CCX) division at Broadcom Inc. He leads a team of architects. In his role, he is responsible for the product architecture definition of Ethernet NIC products, NIC integration... Read More →


Friday March 15, 2019 2:30pm - 2:55pm PDT
Marriott Salon I&II

2:30pm PDT

DCTCP in the OCP Datacenter: Utilizing OCP Hardware to its Maximum Potential by Reducing Tail Latency
Over the past year much work has been done to tie in the DCTCP congestion control algorithm with OCP networking [1] and server [2] hardware. This has helped solve numerous problems related to tail latencies. This talk will describe a method to utilize this work in adopting disaggregated flash storage over TCP within a low power OCP environment focused on the Wedge100 switch and the Yosemite v2 multi-host platform. It will describe the OCP specific operational challenges faced, solutions found, and the impact this has with respect to disaggregated data center operation. Finally, the interaction between the bottlenecks addressed in the Wedge100 rack switches and in the Yosemite platforms will be explored, including how through the combination of queueing structures in the switches and OCP v2.0 mezzanine cards, tail latencies can be reliably capped. Scope: DC network switches, disaggregated flash workloads, storage microbenchmarks, latency sensitive DC apps, and NIC dataplane behavior. 1. https://linuxplumbersconf.org/event/2/contributions/95/ 2. https://linuxplumbersconf.org/event/2/contributions/112/

Speakers
LB

Lawrence Brakmo

Kernel Software Engineer, Facebook
AT

Amal Tariq

Network Hardware Engineer, Facebook
avatar for Nic Viljoen

Nic Viljoen

Associate Director of Engineering, Netronome
Nic works on upstream solutions using SmartNICs, focusing on the interaction between SmartNICs and the Linux kernel using the upstream BPF JIT for the Netronome NFP processor.


Friday March 15, 2019 2:30pm - 2:55pm PDT
211 CD

2:30pm PDT

Live Demo of Zero Touch Deployment
Live demo of installing NOS and orchestration agent on an OCP switch and live rollout of configuration of fully functional network with E-BGP, NAT, Load Balancer, ACLs.


Speakers
avatar for Alex Saroyan

Alex Saroyan

CEO, Netris
CEO of Netris - the new automatic netops platform that turns your physical network into service, and provides network services to your Kubernetes cluster, compute, and storage.


Friday March 15, 2019 2:30pm - 2:55pm PDT
210 DH

2:30pm PDT

Open Source Firmware Automatized Testing
Firmware complexity has increased dramatically during the past decade, while testing features and automatization are still a critical missing part. We will during this talk update the community on the CI systems that we currently develop to support the linuxboot project, with some live demo of the infratructure. We will introduce d some ideas on how to extend the technology to other firmware project like OpenBMC or u-bmc.

Speakers
JV

Julien Viard de Galbert

Embedded and Software Engineer
avatar for Jean-Marie Verdun

Jean-Marie Verdun

CTO, ITRenew


Friday March 15, 2019 2:30pm - 2:55pm PDT
Marriott Salon III

2:30pm PDT

Converged Open Rack Frame Specification Preview
Facebook will be showing an overview of the next generation of their version of Open Rack frame.  The goal for Open Rack V3 is to provide a flexible frame and power infrastructure capable of supporting a wide range of solutions from across the OCP community.

Speakers
SM

Steve Mills

Facebook


Friday March 15, 2019 2:30pm - 2:55pm PDT
210 BF

2:30pm PDT

Datacenter-ready Secure Control Module and Interface (DC-SCM & DC-SCI) - A Modular Building Block Architecture (MBA, the Catalyst)
This Talk explores a modular approach for streamlining hardware security and management by design for compatibility and ease of design, deployment, and debug.

Speakers
avatar for Siamak Tavallaei

Siamak Tavallaei

Principal Architect, Microsoft Azure; CXL Consortium™ Technical Task Force Co-Chair, Microsoft
Siamak Tavallei is a principal architect and technologist instrumental in development and evolution of innovative multi-node, and multi-processor servers as well as technology initiatives in areas of storage/memory hierarchy and heterogeneous, distributed computing. He is a customer-driven... Read More →


Friday March 15, 2019 2:30pm - 2:55pm PDT
210 CG

2:30pm PDT

Unlocking SONiC's Potential for Intent-Based Data Centers: Three Easy Steps
Remove the barriers of deploying, monitoring and managing a data center network, while improving application reliability and availability and security -- all while lowering operating costs. To build an Intent-Based Data Center that automates tasks like configuration of networking devices, you need to provide a cloud-like experience and simplify operations. Learn how to achieve these goals with Intent-Based Networking technology, analytics and telemetry. SONiC can play an integral role in the process to further reduce complexity, provide full lifecycle support (design, deploy, operate) and data gathering with on-box or off-box telemetry. In three steps, Apstra will share how you can unlock a wider set of opportunities and functionality in your network infrastructure to pursue the autonomous data center using SONiC.

Speakers
NT

Nikos Triantafillis

Member of Technical Staff, Apstra


Friday March 15, 2019 2:30pm - 2:55pm PDT
Marriott Salon IV

2:40pm PDT

MiTAC PCIE Gen4 JBOF – Crystal Lake
System Over view
What’s new
Design Introduction
Question

Speakers
avatar for Eric Kuo

Eric Kuo

Engineering VP, MiTAC Computing Technology Corp.
Eric Kuo, current title as engineering VP of MiTAC Computing Technology Corp.,  has been leading MiTAC’s global engineering team dedicated in product development works including server, storage, networking as well as IaaS software and firmware.MiTAC was established in 1982, with... Read More →


Friday March 15, 2019 2:40pm - 2:58pm PDT
212 AB

2:45pm PDT

openEDGE Ecosystem Status
Update on the latest developments with the openEDGE ecosystem adaptation and engagements with potential suppliers and adopters. This presentation will outline the engagement themes and recognized development areas needing further collaboration within industry.

Speakers
avatar for Mike Moore

Mike Moore

Edge Cloud Platform Product Sales, OpenEdge Sub-Project Lead, Nokia
Over 20 years of R&D experience in both the Wireless and Defense Industries prior to transitioning into a customer facing sales and support role.  Now the Western Hemisphere Regional Product Manager for Nokia’s AirFrame Data Center Portfolio, with a focus on creating highly available... Read More →


Friday March 15, 2019 2:45pm - 2:59pm PDT
212 CD

2:50pm PDT

Accelerate IT Deployments with Rack Ready Pod Based Designs - Presented by Schneider Electric
Fully integrated racks has changed how IT is deployed into data centers. Customers are looking for more streamlined and efficient ways to deploy IT. Traditional infrastructure was optimized for server by server installation and is time consuming and expensive. Learn how pod based approaches can reduce capex by 15% and improve IT deployment speed by 20%.

Speakers
avatar for Brian Ballard

Brian Ballard

Business Development Manager, Schneider Electric
Brian Ballard is a Business Development and Strategy Leader with APC by Schneider Electric and is part of a cross functional team that turns strategy into execution throughout the North American zone. Stop by our booth at D9 to learn more about flexible Pod Containment Systems for... Read More →


Friday March 15, 2019 2:50pm - 3:05pm PDT
Expo Hall 1

3:00pm PDT

SONIC/SAI Support in Juniper DC Switches (SONIC / SAI support in Multi-PFE modular chassis) - POC
In this POC, we will demo installation of SONIC/SAI on a Multi-PFE modular chassis . We will demonstrate how SONIC can now manage a multi-pfe modular chassis using JUNOS SAI app. We will SONIC controlling a multi pfe modular chassis forwarding plane.

Speakers
avatar for Lester Bird

Lester Bird

Distinguished Engineer, Juniper Networks
Lester has worked for 33 years in computer networking. He specializes in network stacks and distributed, fault tolerant systems. He is currently leading the multi-PFE SONiC project at Juniper Networks.
avatar for Kumuthini Ratnasingham

Kumuthini Ratnasingham

Senior Product Line Manager, Juniper networks
Holds a Master's degree in Engineering and holds multiple US Patents. Speak @ various events including OpenStack Summits.


Friday March 15, 2019 3:00pm - 3:12pm PDT
Marriott Salon IV

3:00pm PDT

OCP Data plane Acceleration for Edge Cloud
Data plane acceleration is an essential requirement for building a high performance Telco Cloud. Nokia and Mellanox have teamed up to create an integrated OCP compliant solution supporting "switch on NIC" OVS offloading. Nokia's AirFrame openEDGE servers now includes the Mellanox OCP Compliant ConnectX-5 NIC. The ConnectX-5 NIC has an embedded switch which enables orders of magnitude improvement in throughput for cloud networking protocols such as VXLAN. Nuage Networks has developed an SDN solution which works on this platform enabling automated programming of the NIC and providing VIM integration with OpenStack, K8s, and other cloud management systems. This review session covers the design, performance and benefits (functional and economic) of this solution.

Speakers
avatar for Mike Moore

Mike Moore

Edge Cloud Platform Product Sales, OpenEdge Sub-Project Lead, Nokia
Over 20 years of R&D experience in both the Wireless and Defense Industries prior to transitioning into a customer facing sales and support role.  Now the Western Hemisphere Regional Product Manager for Nokia’s AirFrame Data Center Portfolio, with a focus on creating highly available... Read More →
AB

Ash Bhalgat

Sr. Marketing Director, Mellanox Technologies
Ash Bhalgat is an accomplished hi-tech industry leader focussed on the intersection of technology, products and markets. He has passion for leading innovative and dyanmic technology organizations to drive vision, strategy, execution and marketing of hardware, software and cloud products... Read More →
MI

Mark Iskra

TME, Nuage Networks


Friday March 15, 2019 3:00pm - 3:14pm PDT
212 CD

3:00pm PDT

HW/SW Co-design for Predictable IO Latency
While the industry has been focusing on improving the IOPS & density of storage, the growth rate hasn’t been the same for both across HDD’s, Flash & SCM (storage class memory). In order to keep up with the required IOPS/TB, applications pay the cost with the latency credits. Using a RocksDB use-case, we discuss about the various considerations & trade-offs in the HW/SW co-design that influences an optimal solution that solves for predictable IO latency.

Speakers
NK

Navneeth Kankani

HW solutions, Facebook


Friday March 15, 2019 3:00pm - 3:18pm PDT
212 AB

3:00pm PDT

Eco-system Enabling of Liquid Cooling Ingredients
The continued demand for increased compute performance comes with an increase in server power and power density. The increased server power requires more efficient cooling capabilities compared to traditionally used air cooling. Direct liquid cooling using cold plates can provide the needed cooling capability, as well as fit within a small volume to allow for greater server and compute density in the rack. Even though liquid cooling is not a new cooling strategy for servers, the industry is not ready to efficiently deploy liquid cooling solutions in mass. Therefore, Intel is working to enable liquid cooling ingredients with industry partners. The intent is to create more available, lower risk, and faster time to market opportunity for OEMs to deploy liquid cooling solutions. Intel’s efforts are specifically focused on manifold distributed liquid cooling at the server rack level (i.e. the secondary cooling loop). The secondary loop is the fluid loop providing the cooling liquid from the Cooling Distribution Unit (CDU) through the rack manifold to the server cold plates, and then back again to the CDU through the rack manifold. The liquid cooling ingredients in the secondary loop included in Intel’s enabling efforts are: cooling liquid, CDU, interchangeable Universal Quick Disconnects (UQD), cold plates, and design considerations for the secondary loop including the in-rack manifold. Intel’s enabling efforts, together with Intel’s strategy to enable these ingredients will be discussed in the presentation at the OCP summit. This presentation will not discuss specific internal chassis fluid loops, or cold plate designs, except for the CPU cold plate.

Speakers
avatar for Mark Sprenger

Mark Sprenger

Tech Lead Engineer, Intel
Mark Sprenger is Senior Mechanical Engineer and Technical Lead in Data Platform Group responsible for external development of cooling solutions for commercial datacenter products. Mark has over 25 years of industry experience in advanced development covering a wide variety of technologies... Read More →
avatar for Jessica Gullbrand

Jessica Gullbrand

Tech Lead Thermal Engineer, Intel
Jessica is a Thermal Tech Lead Engineer at Intel in the Data Platform Group (DPG). She is currently working on energy efficiency, immersion, and liquid cooling cold plate technologies. Jessica has worked on new and innovative technology solutions in fluids, acoustics, and thermals... Read More →


Friday March 15, 2019 3:00pm - 3:25pm PDT
210 E

3:00pm PDT

From PUE to ERE to Synergy
This session is aimed at positioning datacenter concepts which are fully focused on synergy by maximizing exergy and embracing the advantages for this.
This session reviews and considers future datacenters as part of a wider industrial business environment, within the four utility domains
  1. power
  2. IT/data
  3. heat
  4. CO2
and distribution of all four above, within the datacenter context.

Speakers
avatar for Wim Buters

Wim Buters

Principal Commercial Officer, Asperitas
Seasoned professional on business-to-business delivery. Accountable for further expansion of Asperitas' commercial position on the market. Asperitas being developer, supplier and producer of high-end solutions for the data centre industry based on immersed compute technology.



Friday March 15, 2019 3:00pm - 3:25pm PDT
211 AB

3:00pm PDT

Next Generation Intel Xeon Scalable Processors for Machine Learning
Intel is accelerating AI performance on Intel Xeon Scalable Processors with hardware features and software optimizations boosting performance across a wide range of AI/DL/ML workloads. Starting with the codename Cascade Lake Intel Xeon processors, Intel added the Vector Neural Network Instruction (VNNI) to the AVX-512 instruction set to further boost AI inference performance. In this session, we will provide a detailed overview of VNNI and the optimizations Intel is doing across popular frameworks. And, we’ll live demo these benefits using publicly available frameworks.

Speakers
avatar for Shivani Sud

Shivani Sud

Architect, Intel
JL

Jianhui Li

Principal Engineer, Intel
avatar for Andres Rodriguez

Andres Rodriguez

Sr. Principal Engineer, Intel
Andres Rodriguez is a Sr. Principal Engineer in the Data Center Group (DCG) where he designs deep learning solutions for Intel’s customers and provides technical leadership across Intel for deep learning products. He has 15 years of experience working in artificial intelligence... Read More →
NS

Niveditha Sundaram

Director of Engineering, Intel


Friday March 15, 2019 3:00pm - 3:25pm PDT
210 A

3:00pm PDT

OCP Open Rack Manager Controller Subproject
In 2018, OCP established a project for System Firmware project to deliver open source BIOS and supported the unification of the baseboard management controller firmware in the OpenBMC repository. OCP has identified the rack manager controller firmware as the next opportunity for manageability interface specification and an open source repository. Hence, it has launched the OpenRMC subproject. This presentation will describe the goals and deliverables of the OpenRMC subproject. It will also describe the current plans and status.

Speakers
avatar for John Leung

John Leung

Principal Engineer, Intel
John Leung is a Principal Engineer at Intel Corporation in the Data Platforms Group. John has been in the computer industry for over 36 years: architecting, designing and coding. At Intel, John is currently focused on the manageability interfaces for data center compute platforms... Read More →
AL

Alfie Lew

Inspur Systems


Friday March 15, 2019 3:00pm - 3:25pm PDT
Marriott Salon I&II

3:00pm PDT

LinuxBoot: Boot anything from Linux
With LinuxBoot making its way into data centers, people are starting to wonder: can I boot something other than Linux? Of course you can! We'll talk about getting multiboot kernels such as Xen or VMware ESXi or Akaros to boot from Linux, as well as goals such as getting Windows to boot.

Speakers
avatar for Chris Koch

Chris Koch

Software Engineer, Google
Chris is a software engineer at Google and currently works on bringing LinuxBoot to production at Google. In a previous life, he worked on various kernel features for gVisor and supported gVisor customers like AppEngine.


Friday March 15, 2019 3:00pm - 3:25pm PDT
Marriott Salon III

3:00pm PDT

Finite Element Modeling of the 12V Power Distribution in the Open Rack Standard 2.0. Analysis of Losses and Temperature Rise
It is still unclear within the OCP community what is the maximum current and power limit that could be delivered with the 12V busbars defined in the Open Rack Standard 2.0. Bel Power Solution, Rittal and Circle B presents a study to assess losses, voltage drop, airflow and temperature rise in different configurations reflecting existing or potential applications in the market. The study reflects the expertise of the three companies and will constitute a guideline for any company that decides to adopt an Open Rack solution. The study will address different rack configurations and IT loading. The results show the performance of a 12V rack power distribution with the current busbars and demonstrates better performance with an improved busbar designs to cover more challenging applications.

Speakers
SB

Samuel Brumhard

Product Manager, Rittal
NC

Nicola Cinagrossi

Engineering Director, Bel Power Solutions and Protection
avatar for Menno Kortekaas

Menno Kortekaas

Technical Director, Circle B
Providing OCP based solutions to European datacenters


Friday March 15, 2019 3:00pm - 3:25pm PDT
210 BF

3:00pm PDT

Enabling the Next-Generation of SDN
This talk will provide an update on Stratum -- a next generation, thin switch OS that provides silicon and hardware independence -- that was first introduced at the 2018 OCP Summit. Stratum adds new SDN-ready interfaces from the P4 and OpenConfig communities to ONL that enable programmable switching chips (ASICs, FPGAs, etc.) and traditional switching ASICs alike. We will describe how Stratum can leverage SAI’s programmable model (and flexible P4 architecture) for consistent control and management and explain how Stratum can help enable enhanced data plane programmability and telemetry for control planes, including SONiC. Finally, this talk will cover how the community has generalized Google’s seed contribution for additional whitebox targets, and demonstrate Stratum on a fabric of OCP devices controlled by an open source control plane.

Speakers
AG

Alireza Ghaffarkhah

Staff SWE, Google
Alireza is a Staff Software Engineer working in Platforms team at Google.
DG

Devjit Gopalpur

Software Engineer, Google
Devjit is a software engineer at Google and is part of the core development team that developed and hardened early versions of Stratum. He continues to contribute to project and is a leader within the open source community.
SN

Shashank Neelam

Senior SWE, Google
BO

Brian O'Connor

Stratum Lead and MTS, Open Networking Foundation (ONF)


Friday March 15, 2019 3:00pm - 3:30pm PDT
211 CD

3:00pm PDT

Architecture and Design of Open Network Linux
"This engineering workshop will focus on the architecture and design of Open Network Linux. The workshop will provide details on the system and platform integration and Open Network Linux Platform APIs. A demonstration of ONL features will be provided along with a walkthrough on how to add new platforms."

Speakers
JT

Jeffrey Townsend

Vice President, Engineering, Big Switch Networks


Friday March 15, 2019 3:00pm - 3:30pm PDT
210 DH

3:00pm PDT

Gen-Z Technology: Enabling Memory Centric Architecture
Speakers
avatar for Greg Casey

Greg Casey

Dell CTO - Server Strategist, GenZ - Dell/EMC
Greg Casey is a Senior Architect and Strategist in the Server CTO team of Dell/EMC and is a Board Member of the PCI-SIG. At Dell EMC, he has been involved with the Definition, Development and Design of Modular Server Systems. Prior to joining Dell in 2006, Greg was with QLogic... Read More →


Friday March 15, 2019 3:00pm - 3:30pm PDT
210 CG

3:10pm PDT

MiTAC’s Offering for OCP and the Next Gen Infrastructure - Presented by MiTAC
  • MiTAC’s introduction and offering for the infrastructure
  • MiTAC’s status and future plan for OCP solution
  • MiTAC’s OCP solution offering for VCO

Speakers
avatar for Alfred Chao

Alfred Chao

Special assistant to President Office, MiTAC


Friday March 15, 2019 3:10pm - 3:25pm PDT
Expo Hall 1

3:15pm PDT

cRPD – Cloud-Grade Routing as a Micro-Service for Open Networking Platforms
Building web scale data centers require simple, open, scale-out, secure, automated software architectures. This requires disaggregation of the technology stack so that applications can run in any cloud, cloud workloads can run on any device, and software is not locked to hardware. Here we will present a solution using JUNOS control plane (cRPD) running as a software micro-service which provides management, routing, programmability and rich telemetry in a container package. Will demonstrate following use cases using cRPD, 1. Scale out Route Reflector and Route Server with Kubernetes 2. Routing on the host (BGP, ISIS, OSPF, BFD, Netconf, OpenConfig, gRPC based APIs & telemetry) 3. Egress peer engineering in Data Centers 4. Building Overlays in Data Centers (L3VPN, EVPN) with MPLSoUDP, VxLAN, SR-MPLS 5. Securing host in data centers - filtering and DDoS protection using BGP Flow spec 6. Routing on SONiC for white box switches 7. Routing on edge cloud gateway

Speakers
MG

Manish Gupta

Engineering Director, Juniper Networks
VN

Vinay Nallamothu

Software Engineer, Juniper Networks


Friday March 15, 2019 3:15pm - 3:27pm PDT
Marriott Salon IV

3:15pm PDT

DO-160 and Twinlake Servers
The Open Edge workgroup has introduced the need for wider range temperature servers addressing the IoT and edge computing requirements. These requirements are in some ways close to the DO160 standard which is a reference in aerospace industry. After a short introduction of this standard, we will present during this talk a first status and overview of potential adaptions required to reach DO160 compliancy by focusing our work on the twinlake server.

Speakers
avatar for Nicolas Roels

Nicolas Roels

HW Technology Director, Thales
avatar for Jean-Marie Verdun

Jean-Marie Verdun

CTO, ITRenew


Friday March 15, 2019 3:15pm - 3:29pm PDT
212 CD

3:20pm PDT

HDD IOPS : The Never Ending Story
In the never-ending search for higher density, hard disk drives have continued to create a widening performance gap in IOPS/TB. Over time software has evolved to compensate for this by optimizing hot and cool data distribution, as well as multiple layers of caching. With SSDs continuing to reduce in price, how can HDDs improve and be effectively used to bridge the performance gap? Multi-Actuator drives are on their way, and their introduction will require us to consider the impact to our storage platforms and the TCO models to see if they make sense. With higher performance comes higher power and thermal challenges.

Speakers
avatar for Jason Adrian

Jason Adrian

Storage Hardware Architect, Microsoft Azure
Jason is passionate about hyperscale and scale-out storage systems and storage architectures. His experience ranges from high-performance flash systems to archival systems, and everything in between.


Friday March 15, 2019 3:20pm - 3:38pm PDT
212 AB

3:30pm PDT

The New Networking Edge: OSS Software and Hardware Enabling 5G & IOT
This session will explore how open source hardware and software are working together to support emerging network edge use cases, such as 5G and IoT. This includes work being done among correlating open source communities and ecosystems.

Speakers
HK

Heather Kirksey

VP, Community and Ecosystem Development, Networking, The Linux Foundation


Friday March 15, 2019 3:30pm - 3:44pm PDT
212 CD

3:30pm PDT

Open19 Technology - Presented by LinkedIn
The Open19 platform is the next generation technology of choice for data center and Edge for modular modernized data center infrastructure and server build outs. Mr. Zhuk will give a deep dive of the Open19 platform, including technology and business aspects of Open19.

Speakers
SZ

Sergiy Zhuk

Director, Hardware Engineering, LinkedIn


Friday March 15, 2019 3:30pm - 3:45pm PDT
Expo Hall 1

3:30pm PDT

PUE 0.7 - Data Center of the Future
The technology exists today for practical Data Center Heat Recovery. This session discusses practical heat recovery design and application scenarios that could lead to a "Heat Recovery Adjusted PUE of 0.7". These solutions would significantly reduce the consumption of fossil fuels in public and commercial facilities. The convergence of IOT, Edge Data Centers, Cloud Data Centers, and corporate sustainability goals would benefit greatly from these solutions.

Speakers
avatar for Michael Beatty

Michael Beatty

Data Center Facilities Consultant, Data Center Dimensions
Mike is passionate about helping his clients improve their Data Center Facilities Operations and Reliability. He is a champion for new Data Centers that actually recover their heat for useful purposes, like heating schools and other public and commercial buildings. He is a certified... Read More →


Friday March 15, 2019 3:30pm - 3:55pm PDT
211 AB

3:30pm PDT

OCP Accelerator Module (OAM)
The OCP Accelerator Mezzanine Module design specification defines the form factor and common specifications for a compute accelerator module, a compliant base board design enabling interoperability across multiple ASIC or GPU based mezzanine modules, and a based board design plan. The mezzanine module form factor facilitates scalability across accelerators by simplifying the system solution when interconnecting communication links among modules in comparison with a PCIe Add-in card form factor.

Speakers
avatar for Siamak Tavallaei

Siamak Tavallaei

Principal Architect, Microsoft Azure; CXL Consortium™ Technical Task Force Co-Chair, Microsoft
Siamak Tavallei is a principal architect and technologist instrumental in development and evolution of innovative multi-node, and multi-processor servers as well as technology initiatives in areas of storage/memory hierarchy and heterogeneous, distributed computing. He is a customer-driven... Read More →
avatar for Whitney Zhao

Whitney Zhao

Engineer, Facebook
Whitney is a seasoned hardware engineer leading AI/ML system design in Facebook. Whitney has led multiple hardware generations ranging from general purpose 2S system such as Tioga Pass to ML JBOG Big Basin systems, all of which have been contributed to OCP. She has been driving multiple... Read More →
avatar for Tiffany Jin

Tiffany Jin

Mechanical Engineer, Facebook
avatar for Richard Ding

Richard Ding

Principal AI System Architect, Baidu
Richard Ding is Principal AI System Architect from Baidu Infrastructure Team. He leads architecture design of Baidu AI Computing Platform KongMing including AI computer X-MAN, high-performance parallel file system FAST-F, large scale and distributed communication engine Halo, as well... Read More →


Friday March 15, 2019 3:30pm - 3:55pm PDT
210 A

3:30pm PDT

Open Cloud Server Management in the Exchange Online Service
Microsoft's Exchange Online service has converged on the Open Cloud Server (OCS) and Project Olympus platforms for its server infrastructure. We are operating a fleet of 100s of 1,000s of OCP servers, with varying generations and SKUs of hardware, in over 100 datacenters worldwide. This session will cover how we discover, provision, monitor, diagnose, and repair our OCP servers, how we have engineered solutions to current OCP Redfish limitations, and discuss opportunities for the Hardware Management group to improve this experience.

Speakers
MM

Matt McCormick

Senior Program Manager, Microsoft


Friday March 15, 2019 3:30pm - 3:55pm PDT
Marriott Salon I&II

3:30pm PDT

Improving Cloud System Up-Time with Run-time Firmware Update and Activation
Today when the system firmware are updated, it is typically requires system reset and reload of the OS is required to get the services back up and running again. This means, the system services need to be migrated or restarted to accommodate for new firmware. The cloud workloads often demands for uninterrupted services from compute systems. The session would outline vision for run-time firmware and run-time firmware activation for Open Compute Project platforms. The session will revisit various typically used firmware components and various pre-requests need to be in place for successful run-time firmware update and activation for Open System Firmware.

Speakers
avatar for Mallik Bulusu

Mallik Bulusu

Principal Firmware Engineering Manager, Microsoft Corporation
Mallik Bulusu is a Principal Firmware Engineering Manager (Azure) at Microsoft Corporation. He has been specializing in design and development of system firmware using UEFI paradigm since 1999. Mallik holds 65+ patents on novel advances in system firmware design for Server and Mobile... Read More →
avatar for Murugasamy Nachimuthu

Murugasamy Nachimuthu

Sr. Principal Engineer, Intel Corporation
Murugasamy (Sammy) Nachimuthu is a lead cloud solutions architect, focusing on improving cloud service provider infrastructure. Sammy joined Intel in 1995, he played key role in architecture of Rack Scale Design (RSD) architecture, Intel® Optane™ DC Persistent Memory, PCIe, Memory... Read More →


Friday March 15, 2019 3:30pm - 3:55pm PDT
Marriott Salon III

3:30pm PDT

Challenges and Solutions in Multi Vendor Inter op
This session will high light electrical and Mechanical challenges in Multi vendor Inter op and suggest solutions which will provide a building block for Inter op specifications for various Power components including PSU's and BBU's. This session will provide an update on additional work that has been completed building upon previous work which was presented in Amsterdam

Speakers
avatar for Harry Soin

Harry Soin

Senior Director Technical Marketing, Artesyn Embedded Power
Harry Soin is the Senior Director of Technical Marketing at Artesyn Embedded Power and is responsible for the Hyperscale segment. Harry has been working in this capacity since Jan 2016 in San Jose,CA. Prior to this Harry managed the China Design Engineering facility for Artesyn which... Read More →


Friday March 15, 2019 3:30pm - 3:55pm PDT
210 BF

3:30pm PDT

SONiC Deployments Powered by Programmable Dataplane
Since the initial announcement at 2016 OCP Summit, SONiC’s popularity has increased tremendously. An expanded feature set and a diverse set of trials and deployments were made possible through a vibrant open source community. However, the potential of open source networking software such as SONiC has not been fully realized before the introduction of programmable silicon. In this session, we will discuss real world SONiC trials and deployments on top of programmable networking silicon. We will focus on network designs and use-cases around telemetry, distributed storage and scalable tunneling.

Speakers
avatar for Arkadiy Shapiro

Arkadiy Shapiro

Product Line Manager, Barefoot Networks, an Intel Company
Arkadiy Shapiro is a Product Line Manager at Barefoot Networks, an Intel company. In this role, he is responsible for product strategy and marketing of Software Development Environment that enables customers to create programmable network solutions with Barefoot’s ASICs. In the... Read More →


Friday March 15, 2019 3:30pm - 3:55pm PDT
Marriott Salon IV

3:30pm PDT

In-depth Discussion: ACS Immersion Standards and Best Practices
Speakers
avatar for Rolf Brink

Rolf Brink

CEO, Asperitas
Product development for cloud architectures and datacentre infrastructure, combined with a passion for cleantech innovations like liquid cooling, are the foundation of the developments which drive Asperitas.Asperitas is a clean-tech company focused on greening the datacentre industry... Read More →



Friday March 15, 2019 3:30pm - 5:00pm PDT
210 E

3:40pm PDT

Learning of Designing Project Olympus JBOF with EDSFF SSD
In the presentation, Wiwynn will share the lessons learned in developing Project Olympus FX-16 system which is equipped up to 16 pieces 1U Long with 18mm thick EDSFF NVMe SSDs (SFF-TA-1007). Considering the high density and serviceability requirements, lots of trade-offs have to be made during the design process, such as SSD form factor definition, blind mating, balance of airflow impedance & thermal efficiency. And we will also share design tips on control and protection mechanism for hot-plugging, SMbus and supporting dual-port feature, etc. With extreme high storage capacity and hot plug serviceability, the design allows a low maintenance cost for data center operation. The power consumption results have confirmed our efficiency of the thermal design with superior thermal efficiency of low CFM per Wattage, which will reduce data centers OPEX.

Speakers
avatar for Mark Shaw

Mark Shaw

GM Advanced Hardware Dev, Microsoft
Standardized, scalable hardware for the Cloud! Open CloudServers!Mark Shaw is an Architect and developer of Server Hardware for Microsoft’s Cloud+Enterprise Division. Mark and his team are working with the industry on developing the next set of server hardware.
AC

Antson Chang

Assistant Technical Director, Wiwynn Corporation


Friday March 15, 2019 3:40pm - 3:58pm PDT
212 AB

4:00pm PDT

Leverage OCP Design Advantages on EIA 19” GPU Server
For high power consumption systems such as GPU server, the thermal solution is a big challenge. The breakthrough design of isolated cooling air channels is adopted in server board, GPU subsystem, and power supply units of Wiwynn SV500G3. The architecture sets the deep learning performance benefits free from thermal barriers under eight high power computing accelerators. Cooling efficiency is optimized with well managed air flow and hence achieved excellent CFM per wattage ratio. SV500G3 is also able to operate at higher ambient temperature for better PUE. The server board leverages OCP Project Olympus server board which had been massively used and verified in hyper-scale data centers, that guarantees availability. Compared to traditional GPU servers, the system is front-accessible with tool-less maintenance. Rack cabling and maintenance efforts are reduced to a minimum, enabling our customers to focus more on rack configuration and deployment scalability.

Speakers
GL

Gregary Liu

Director, Wiwynn
Server- and Rack-level design for the cloud services the customer would like to address!


Friday March 15, 2019 4:00pm - 4:25pm PDT
210 A

4:00pm PDT

CHIPSEC on non-UEFI Platforms
CHIPSEC is a firmware threat assessment tool used to help verify that systems meet basic security best practices. The tool’s threat model is primarily based on Unified Extensible Firmware Interface (UEFI). However, other firmware may have different threat models that will cause failures in different CHIPSEC modules. This session is a brief overview of CHIPSEC, limitations of the tool, failures seen on different types of firmware, and information on developing new test modules.

Speakers
avatar for Stephano Cetola

Stephano Cetola

Open Source Program Manager, Intel
Stephano is an Open Source Program Manager at Intel contributing to TianoCore and CHIPSEC. His main focus is on community engagement and improving the developer experience. In his free time he enjoys exploring the thermal tolerances of various electronic circuits.


Friday March 15, 2019 4:00pm - 4:25pm PDT
Marriott Salon III

4:00pm PDT

SONiC+: Adding Application Logic to Open NOS
Now that SONiC is in production we should start enjoying the fruits of having an open Network OS. The motivation of an open NOS doesn't stop with having an open source BGP with a configuration DB and bunch of CLIs, it's about what logic can we put in the network, now that we control the NOS. SONiC+ is exactly about it. Taking advantage of flex SAI interface, Mellanox is introducing set of containerized applications running in SONiC that offloads traditional appliance functionality to a COTS SONiC based switch

Speakers
MK

Matty Kadosh

Principle Architect, Mellanox Technologies
avatar for Aviad Raveh

Aviad Raveh

VP, BD technology, Mellanox Technologies
Aviad is the VP, BD technology, responsible of driving the Ethernet switch business development. Aviad collaborates with the biggest T1 and T2 cloud and web companies in the US and China. Until recently, he led the Ethernet switch software architecture and was responsible for building... Read More →


Friday March 15, 2019 4:00pm - 4:25pm PDT
Marriott Salon IV

4:00pm PDT

OCP Equipment for Everyone
Today OCP gears is mainly used by hyperscalers. Testing and buying OCP equipment require acquiring at minimum one rack and it could slow down the spread to smaller companies or research lab (private or university). In open source software communities, we can find a lot of students, researchers and engineers from small companies who enrich projects by new ideas which is missing within OCP community. Objective of our project is to accelerate OCP equipment adoption by those users which could bring more recognition and thereby increase OCP community. We would like to provide a mini rack with basically 4x « 1/3 rack » form factor OCP equipment (open rack V1 and V2) in a new chassis with one power supply and one 1U switch at an affordable price (

Speakers
JC

Jean-Jacques Chanut

Industrial Designer, ITRenew
avatar for Jean-Marie Verdun

Jean-Marie Verdun

CTO, ITRenew
TV

Thomas Van Liefde

Electronical engineer, ITRenew


Friday March 15, 2019 4:00pm - 4:30pm PDT
211 AB

4:00pm PDT

Power Capping with Cloud Super Apps Performance SLA
We developed and deployed an advanced power capping method on Alibaba datacenter infrastructure that can meet pre-defined performance SLA for cloud super apps. Our datacenter manager platform exposes only power capping interface to cloud super apps, and under the hood our power master and power agents evaluate and choose the best power limiting and budgeting control knobs and parameters, and achieve the best trade-off between power and performance, and meet guaranteed performance SLA. Our datacenter manager works on commodity and customized servers, non-OCP and OCP both OK. We'll present our architecture and key technology break-throughs, as well as how we take advantage of OpenBMC.

Speakers
NA

Nishi Ahuja

Senior Principal Engineer, Intel
JS

Jun Song

Chief Power Architect, Alibaba Group
Jun (Justin) Song is chief power architect of Alibaba Group responsible for server and data center power optimization and management. Before joining Alibaba Justin worked as a silicon, chip and platform architect with Intel for 14 years. Justin made technical presentations on public... Read More →
ES

Edmund Song

Software Architect, Intel Corporation


Friday March 15, 2019 4:00pm - 4:30pm PDT
Marriott Salon I&II

4:00pm PDT

Power Shelf Input Power Connector Design Workshop
The Open Rack Standard is driven off of open, non-proprietary designs that can be developed cheaply and easily. This philosophy has been used to drive down costs in many areas except for one: the electrical connector. This session will highlight the progress made in developing a Universal Connector for input power to the power shelf, and to solicit feedback from the community before a standard is finalized

Speakers
avatar for Will Stewart

Will Stewart

Data Center Segment Manager, Harting


Friday March 15, 2019 4:00pm - 4:30pm PDT
210 BF

4:00pm PDT

Panel Discussion on Computational Storage
Speakers
avatar for Jorge Campello

Jorge Campello

Senior Director, Western Digital
avatar for Jason Adrian

Jason Adrian

Storage Hardware Architect, Microsoft Azure
Jason is passionate about hyperscale and scale-out storage systems and storage architectures. His experience ranges from high-performance flash systems to archival systems, and everything in between.
avatar for Scott Shadley

Scott Shadley

Storage Technologist, VP Marketing, Co-Chair SNIA Computational Storage TWG, NGD Systems
Doing my best to improve the world with NVMe Computational Storage Solutions from NGD Systems. Lover of Technology, Innovation and a bit of a Socialite.If it is Storage, I dabble...
avatar for Yahya H. Mirza

Yahya H. Mirza

YCEO/CTO, Aclectic Systems Inc.
Yahya H. Mirza’s original background was aeronautical engineering and he was initially employed by Battelle Research Labs. His experience at NASA Ames in 1989, simulating the hypersonic aerodynamics of waveriders on the Cray- YMP, brought him to an important realization. A clear... Read More →
avatar for Thad Omura

Thad Omura

EVP Marketing & Operations, ScaleFlux
I enjoy sharing and hearing about new applications for Computational Storage, how it is being deployed today in volume, and the different architectural models that impact various workloads in database, big data, CDN and AI/ML applications.
avatar for Stephen Bates

Stephen Bates

CTO, Eideticom
Doing amazing things with NVMe Computational Storage!
SN

Shahar Noy

Sr Director Product Marketing, Marvell
Shahar Noy is a Senior Director of product marketing for Data Center Storage Solutions at Marvell. Shahar joined Marvell’s superhero family in 2017 and other than occasionally saving the world he is busy resolving some of the biggest storage challenges in modern data centers. With... Read More →


Friday March 15, 2019 4:00pm - 4:55pm PDT
212 AB

4:30pm PDT

OCP AI Accelerator Module for Deep Learning Training with Intel Nervana Neural Network Processor(NNP)
Introduce the OCP form factor AI accelerator module for deep learning training with Intel Nervana Neural Network Processor. Discuss detail specifications and system level design guide, including multi-card interconnect topology designs to enable superior scaling performance.

Speakers
avatar for Hang Song Kok

Hang Song Kok

Principal Engineer, Intel


Friday March 15, 2019 4:30pm - 5:00pm PDT
210 A

4:30pm PDT

Common Firmware DRAM Training Data Interface
Currently there is no standard way to export memory related training data from the firmware. Kernel and other monitoring software need to rely on platform specific drivers to get this data. In this presentation we propose a methodology for UEFI based and other Open Firmware to export training data for DDR4/DDR5 to higher layers of software. The proposed methodology addresses both PEI only architecture as well as complete UEFI stack. The proposal defines the content of ACPI (Advanced Configuration and Power Interface) BIOS ACPI Data Table (BDAT) table.

Speakers
avatar for Natarajan V Sivagar

Natarajan V Sivagar

Software Engineer, Google Inc


Friday March 15, 2019 4:30pm - 5:00pm PDT
Marriott Salon III

4:30pm PDT

SONiC + ONL: Open-Source NOS with Broader Platform Support
Today SONiC mainly supports a subset of OCP networking devices specific to the needs of the hyper-scale community. Using Open Network Linux’s platform layer in SONiC increases the reach of SONiC by providing more hardware options, required by a broader set of organizations. Any networking device supported in ONL using a networking chip supported by SAI can be supported for SONiC. For example as demonstrated at the last OCP summit, a combination of SONiC + ONL allowed NTT to demonstrate the Edge-Core Cassini system running SONiC as the data-plane. SONiC + ONL enables SONiC support on wide array of hardware platforms and use cases, thus accelerating its adoption, including SaaS, cloud, managed SP and telco data centers.

Speakers
avatar for Wataru Ishida

Wataru Ishida

Field Application Engineer, NTT Electronics America
avatar for Steven Noble

Steven Noble

Principal SDN Evangelist, Big Switch Networks


Friday March 15, 2019 4:30pm - 5:00pm PDT
Marriott Salon IV