Microsoft Contributes Next Generation Server Design to Open Compute Project
Kushagra VaidGeneral ManagerServer Engineering
Yesterday I gave a keynote address at the Open Compute Project (OCP) European Summit, where I announced the contribution of the 2nd-generation Open CloudServer (OCS v2) specification. In January, Microsoft joined OCP and submitted the first OCS specification, born from our experience in our own cloud-scale datacenters. OCS v2 incorporates a number of improvements, delivering greater performance and more flexibility. In addition, Microsoft Open Tech announced today the release of an open source reference implementation of operational management tools for OCS v2. In all, we at Microsoft are very excited about the tremendous momentum we see in the Open Compute Project, and see it as an excellent community to foster more efficient datacenters and accelerate the adoption of cloud computing.
It’s been great to see tremendous support from industry partners contributing to the OCS v2 design. At the summit, we will show an array of OCS v2 designs and platforms, including from the following partners:
Overall, the OCS v2 specification is the convergence of a number of design points. Meeting these design points in one server is critical for Microsoft to meet its service delivery and datacenter operations goals, and also the needs of the community who’ll use OCS v2 for their own datacenter operations.
The first design point for OCS v2 is to support a diverse range of cloud services. The specifications we’re contributing to OCP reflect our long history in datacenter architecture and cloud computing. To date, Microsoft has invested more than $15 billion in our global cloud infrastructure, and today we provide upwards of 200 cloud services to 1 billion customers and 20 million businesses in more than 90 markets around the world. Over the last six months, the OCS v2 design has been thoroughly tested in our own datacenters, from powering IaaS and PaaS services in Windows Azure, to hosting e-mail and collaboration services in Office 365, to hosting latency-sensitive gaming services in Xbox Live. Converging onto a unified, flexible design allows us to optimize the economics of our supply chain, while delivering a diverse array of cloud services from one underlying server platform.
Our second design point is ensuring OCS v2 can be deployed into any region or colocation facility around the world. OCS v2 takes into account the variation of requirements across geographies, making it simple and efficient to deploy the design in a single region, or as part of a global footprint. OCS v2 enables compliance with differing electrical standards that affect power and voltage to the rack, as well as safety and regulatory certifications that impact weight, rack size, mechanical, and cooling parameters.
The last design point is to optimize performance relative to TCO. OCS v2 delivers the high performance our online services need, while minimizing overall TCO. Reaching the sweet spot for a number of server performance metrics was critical for OCS v2 to be deployed widely across Microsoft’s own global datacenter footprint. OCS v2 introduces numerous new innovations, including:
- A dual-processor design, built on Intel Xeon E5-2600 v3 processors, enabling 28 cores of compute power per blade, and reflecting Microsoft’s joint engineering collaboration with Intel to develop the next generation board.
- Advanced networking for low latency, high bandwidth, highly-virtualized environments, based on 40-gigabit Ethernet networking, with support for routable RDMA over Converged Ethernet (ROCEv2).
- Flexibility incorporated into the core design itself. This allows the integration of a variety of components and add-on cards, including FPGA accelerators, which enables customers to tune their servers for their own unique workloads.
- Low-cost, high-bandwidth, Flash-based memory support, incorporating the latest form factor for m.2 Flash memory. This allows OCS v2-based servers to incorporate higher-capacity SSDs, while ensuring TCO optimization by virtue of using cost-optimized NAND.
- A compact, high-capacity power supply, capable of delivering 1600 watts of power, with a high holdup time of 20 milliseconds.
- Support for high memory configurations, along with flexibility in the amount of memory deployed, by virtue of support for 128GB, 192GB, and 256GB memory capacity configurations.
Overall, we’re very excited about the industry enthusiasm for OCP, and how the organization is focusing the efforts of all the partners in our industry, to accelerate the rate at which innovations in our industry come to market. We look forward to working with the community to help push datacenter innovation forward. You can find more information about OCS v2 at www.opencompute.org.