Last week we offered an inside look at the new server and rack design concepts being developed by Microsoft Global Foundation Services, the business unit that designs and builds the company’s data centers. Microsoft is constantly refining its designs as it deploys cloud computing platforms that require large numbers of servers (as seen in the above photo from an IT-PAC module filled with servers. When you run that server volume and density, small refinements can produce large benefits across these huge computing platforms.
Today we have some additional details from Microsoft in two presentations that provide a deeper dive into the company’s strategy for optimizing its servers and data centers. Here’s an overview and download links:
Cloud Infrastructure: Server Design Strategy (PDF, 4 MB): This presentation prepared by Distinguished Engineer Dileep Bhandarkar provides details on Microsoft’s server purchase scenarios, the specifications for server power supplies, analyses of performance-per-watt-per-dollar at the system level and server total cost of ownership (TCO), a look at the impact of free air cooling on TCO, and an overview of Microsoft’s approach to Solid State Drives (SSDs).
Datacenter Efficiency: Executive Strategy Brief (PDF, 580K): A holistic view of Microsoft’s strategy for improving its data center efficiency, looking at the factors that affect performance at the component, server and system level.
The presentations are the latest in a series of disclosures of designs and strategies from companies building some of the largest data center infrastructures, including he Open Compute Project from Facebook and a series of presentations from Yahoo on its infrastructure. Still to come: a new round of presentations from Google, which is holding its second Data Center Efficiency Summit in Zurich on May 24.