Going over the session program for Data Center Dynamics Chicago on Sept 16, I found the below description of Mike Manos's keynote.
Microsoft’s Chicago Containerized Data Center
Michael Manos, General Manager, Data Center Services - Microsoft
Join this unique Keynote Session, as Microsoft takes us through their Chicago Containerized Datacenter and its C-blox design. Microsoft is embracing containers as the key to building scalable, energy-efficient cloud computing platforms. The company's bold move is an affirmation of the potential for containers to address the most pressing power, cooling and capacity utilization challenges facing data center operators. The Chicago facility is part of the company’s fleet of next-generation data centers being built to support its Live suite of "software plus services" online applications. Microsoft has developed its own specifications that include, for example, configuration for electrical components and the layout of physical servers, for its containers. Those specs make Microsoft’s containers different from anything on the market today, and a potential opportunity for future Microsoft products. The containers, which Microsoft calls C-blox, are largely self-contained and will require very little hands-on maintenance.
The Microsoft Chicago facility is a two floor design where the first floor is a containerized design housing 150 to 220 40’ containers each 1,000 to 1,000 servers. Chicago is large facility with the low end of the ranges yielding 150k serves and the high end running to 440k servers. If you assume 200W/server, the critical load would run between 30MW and 88MW for the half of the data center that is containerized. If you assume a PUE of 1.5, we can estimate the containerized portion of the data center at between 45MW and 132MW total load. It’s a substantial facility.
Container-based data centers allow for better IT reporting all around, users will be able to chart the IT productivity of each unit and get clean statistics on data such as e-mail usage, search queries and any number of other business processes. As the doors are closed, and because of the level of automation in its systems, Microsoft can run them and accept a certain amount of failure over time, which more cost effective to build redundancy and automation into Microsoft’s data center applications and allow some hardware to fail than it would be to physically manage such a large data center. The hands-off approach also means design can be tweaked to allow for maximum cooling and energy efficiency without worrying about how accessible the systems are to human hands. Of course, Microsoft also builds backbones that link power, cooling, and bandwidth among the containers. In the C-blox world, a truck drops off a data center container and then picks it up again in a few years when Microsoft is ready to switch over to new hardware. Administrators will only enter the physical C-blox in the rarest of occasions which will also allow Microsoft to run the entire Northlake facility with a continuous staff of little more than 20 or 30 employees.
I'll be there ready to blog Mike's presentation, but I'll be busy with my own session later.
Fellow press/bloggers you may want to attend the event as well.