OCP Summit 2017
Download as PPTX, PDF3 likes1,795 views
The Open Compute Project (OCP), initiated by Facebook in 2011, aims to share data center product designs among major tech companies to improve efficiency and modularity in hardware. With contributions from organizations like Microsoft, Google, and Apple, OCP focuses on hardware specifications and standards for servers, storage, and networking to enhance scalability and interoperability in data centers. The initiative has resulted in significant innovations in server design and data center operations, exemplified by projects like Microsoft's Project Olympus and Facebook's advancements in AI workloads.
OCP Summit 2017
- 1. Open Compute Project, czyli jak to robią najwięksi Jarosław Sobel @JarekSobel
- 2. • Open Compute Project • OCP Summit 2017 • General • Facebook • Microsoft • ARM solutions • Open Network • Open Rack 48V • Q&A
- 3. Virtualization expert in financial institution Involved in DataCenter, hardware & cloud projects Certified by: Citrix, VMware, Microsoft, NetApp and RedHat In private: blogger, speaker & writer @JarekSobel xenthusiast.com
- 4. The Open Compute Project (OCP) is an organization that shares designs of data center products among companies, including Facebook, Intel, Nokia, Google, Apple, Microsoft, Seagate Technology, Dell, Rackspace, Ericsson, Cisco, Juniper Networks, Goldman Sachs, Fidelity, Lenovo and Bank of America. The initiative was announced in April 2011 by Jonathan Heiliger at Facebook to openly share designs of data center products. The effort came out of a redesign of Facebook's data center in Prineville, Oregon. After two years, with regards to a more modular server design, it was admitted that "the new design is still a long way from live data centers". However, some aspects published were used in the Prineville center to improve the energy efficiency, as measured by the power usage effectiveness index defined by The Green Grid.
- 5. The Open Compute Project Foundation is a 501 non-profit incorporated in the state of Delaware. Corey Bell serves as the Foundation's CEO. Currently there are 7 members who serve on board of directors which is made up of two individual members and five organizational members. Jason Taylor (Facebook) is the Foundation's president and chairman. Frank Frankovsky (formerly of Facebook and past president and chairman) and Andy Bechtolsheim are the two individual members. In addition to Jason Taylor who represents Facebook, other organizations on the Open Compute board of directors include Intel (Jason Waxman), Goldman Sachs (Don Duet), Rackspace (Mark Roenick), and Microsoft (Bill Laing). • On March 11, 2015 Apple, Cisco and Juniper Networks joined the project. • On November 16, 2015 Nokia joined the project. • On February 23, 2016 Lenovo joined the project. • On March 9, 2016 Google joined the project.
- 7. Facebook OCP Prineville DataCenter PUE = 1.06 Typical DataCenter PUE > 1.4
- 9. Mainframe (1970s) Scale Up (1990s) Scale Out (2000s) 2011+ Proprietary Proprietary Proprietary Open Source
- 10. The Open Compute Project's mission is to design and enable the delivery of efficient server, storage and data center hardware designs for scalable computing. All Facebook Data Centers are 100 percent OCP-enabled. They inlcude Prineville (Oregon) Data Center, Forest City Data Center (N.C.), Altoona Data Center (Ia.) and Luleå Data Center (Sweden). Facebook data centers under construction are Fort Worth Data Center and Clonee Data Center (Ireland).
- 13. DataCenter • Data Center Facility Power • Data Center Facility Cooling • IT Space Layout and Design • Data Center Facility Monitoring and Control • Data Center Facilities Operation Server • Open Rack Compatible Chassis and Sleds • Open CloudServer Compatible Chassis and Blades • Micro-servers, Chassis and carriers included ARM,x86 • Components and Peripherals-riser cards, mezzanine I/O, accelerators, networking. • Specification Standards: Electrical Interfaces, Mechanical Interfaces, Manageability, debug, & Test framework. Storage • OpenRack Storage chassis and sleds • Componentsand Peripherals: SAS, SATA,SSD, NVMe, PCIe,USB, networked enabled storage etc. • Promotion & Adoption of OpenBMC • Compatibility with HW MNGT RMC solutions Open Rack • Rack System Architectures such as the following: • 19” EIA Compatible • 21” Open Rack • Rack-Level Power Systems
- 15. Past OCP Summits • March 9-10, 2016, San Jose, CA • March 10-11, 2015, San Jose, CA • October 30-31, 2014, Paris, France • January 28-29, 2014, San Jose, CA • January 16-17, 2013, Santa Clara, CA • May 2-3, 2012, San Antonio, TX • October 27, 2011, New York, NY • June 17, 2011, Palo Alto, CA
- 20. „Big Basin is our next-generation GPU server, featuring a number of design and performance improvements over Big Sur, the first-generation GPU server we announced last year. With Big Basin, we can train machine learning models that are 30 percent larger because of the availability of greater arithmetic throughput and a memory increase from 12 GB to 16 GB. This enables our researchers and engineers to move more quickly in developing increasingly complex AI models that aim to help Facebook further understand text, photos, and videos on our platforms and make better predictions based on this content.” Big Sur with 8 NVIDIA Tesla M40 (2015) Big Basin with 8 NVIDIA Tesla P100 GPU (2017)
- 21. „Bryce Canyon is our new high-density storage platform and provides 20 percent higher hard disk drive (HDD) density than Open Vault. It is designed to support more powerful processors and more memory, and it improves thermal and power efficiency by taking in air underneath the chassis. Our goal was to build a platform that would not only meet our needs today, but also scale to accommodate new modules for future growth. Bryce Canyon provides a powerful disaggregated storage capability with easy scalability compared with other storage platforms.”
- 22. „Tioga Pass is the successor to the Leopard compute platform, and has a dual-socket motherboard, which uses the same 6.5” by 20” form factor and supports single-sided and double-sided designs. Tioga Pass is also Facebook's first dual-CPU server to use OpenBMC after it was introduced with our Mono Lake server last year. The onboard mSATA connector on Leopard has been replaced with an M.2 slot to support M.2 NVMe SSDs. The chassis is also compatible with Open Rack v2. Tioga Pass upgrades the PCIe slot from x24 to x32, which allows for two x16 slots, or one x16 slot and two x8 slots, to make the server more flexible as the head node for both the Big Basin JBOG and Lightning JBOF.” Leopard (2015) Tioga Pass (2017)
- 23. „Yosemite v2 is a refresh of Yosemite, our first-generation multi-node compute platform. The Yosemite v2 chassis supports both Mono Lake as well as the next-generation Twin Lakes 1S server. Unlike Yosemite, the new power design supports hot service — servers can continue to operate and don't need to be powered down when the sled is pulled out of the chassis for components to be serviced. Although Yosemite v2 uses a new 4 OU vCubby chassis design, it is still compatible with Open Rack v2. Each cubby supports four 1S server cards, or two servers plus two device cards. Each of the four servers can connect to either a 50G or 100G multi-host NIC.” Yosemite (v1) with Mono Lake servers Yosemite v2 with Twin Lakes servers
- 24. • „Backpack: The engineering team shared technical details about Backpack, our second-generation modular switch platform: a combination of twelve simple building blocks (switch elements) in an orthogonal direct chassis architecture that provides 100G at the aggregation and spine layers of Facebook's data center Fabrics. • Wedge 100S: A refresh to Wedge 100, our top-of-rack network switch. The new device improves performance with a Broadwell DE microserver, and adds hardware support for enhanced boot security on both the microserver and the BMC. Like all previous Wedge switches, Wedge 100S also runs OpenBMC and FBOSS software. • New software solutions running on Wedge 100: Facebook has collaborated with Microsoft and Canonical to bring SAI and SONiC to the Wedge 100 platform. This represents another important addition to the growing software ecosystem available for Facebook's networking devices.”
- 25. Project Olympus is Microsoft's next generation rack level solution that is open sourced through Open Compute Project. Project Olympus addresses variety of cloud workloads for Microsoft Cloud and was introduced in November 2016. Since inception Project Olympus has attracted a large group of partners from compute silicon providers to ODMs, OEMs and component manufacturers and is becoming the de facto open compute standard for cloud workloads.
- 26. The Project Olympus contribution to OCP will consist the following: • a new Universal motherboard • high availability power supply with included batteries • 1U/2U server chassis • high-density storage expansion • a new Universal rack power distribution unit for global datacenter interoperability • a standards compliant rack management card
- 27. During the OCP US Summit, Qualcomm, Cavium, and Microsoft demonstrated the version of Windows Server ported for our internal use running on ARM-based servers. The Qualcomm demonstration run on the Qualcomm Centriq 2400 ARM server processor, their recently announced 10nm, 48-core server processor with Qulacomm’s most advanced interfaces for memory, network, and peripherals.
- 28. Microsoft announced with NVIDIA and Ingrasys a new industry standard design to accelerate Artificial Intelligence in the next generation cloud. The Project Olympus hyperscale GPU accelerator chassis for AI, also referred to as HGX-1, is designed to support eight of the latest “Pascal” generation NVIDIA GPUs and NVIDIA’s NVLink high speed multi-GPU interconnect technology, and provides high bandwidth interconnectivity for up to 32 GPUs by connecting four HGX-1 together. The HGX-1 AI accelerator provides extreme performance scalability to meet the demanding requirements of fast growing machine learning workloads, and its unique design allows it to be easily adopted into existing datacenters around the world.
- 29. X-Gene™ - World’s First ARMv8 64-bit Server on a Chip® Solution
- 30. White Box Switches idea
- 31. • Canonical announced their official support of Wedge 100 on Ubuntu Core 16, the latest version of their new operating system for cloud and IoT devices. Ubuntu Core supports a number of different network stacks, including FBOSS (Facebook Open Switching System) and SnapRoute as snaps. • Cavium announced a variation of our second-generation top-of-rack switch, Wedge 100C, based on the original Wedge 100 switch specification contributed to the Open Compute Project. • SnapRoute announced the availability of its FlexSwitch software, a micro-services oriented network operating system, on top of the Wedge 100 platform. • Barefoot Networks announced the Wedge 100B series of switches, a new variation of our Wedge 100 platform. There are two switches in this series — a 1RU 32x100GE switch and a 2RU 65x100GE switch. • Cumulus Networks announced that the Cumulus Linux network operating system fully supports Backpack and Wedge 100. • Apstra announced the general availability of the Apstra Operating System™ (AOS)1.1.1, and an integration with Wedge 100. AOS is a vendor-agnostic distributed operating system for the data center network that disaggregates the operational plane from the underlying vendor network operating system and hardware. This is one of the first solutions to provide a simplified way to deploy and operate networks running a variety of operating systems on open switches, as well as more traditional vendor gear.
- 32. New 48V OCP Open Rack v2 • Based upon the Open Rack Standard v2 • Single bus bar – simplified electrical design • Shallow depth (762mm) rack design • N+1 single phase 48V power rectifiers • Stacked N+1 48V Lithium Ion batteries • Ready to deliver 48V directly to CPU Advantages of 48V • Eliminates tight bus voltage requirements • Smaller/Cheaper bus bar & connactors (lower current) • More efficient rectifiers (fewer power step-downs) • No additionals powe converter required for the batteries • Easies to scale up to higher power (24-36KW) • Support up to 60 nodes per rack
- 33. • Intel® Rack Scale Design (Intel® RSD) • High temperature DataCenters • 100G NICs & tranceivers in Facebook DCs • New management environment: RedFish, OpenBMC, etc. • OCP Modular Networking for Telco • SDN, SDS, SDx • Engineering Workshops • Presentations • Meetings • Live discussions • Expo Hall • … and much, much more