Microsoft picks up data centre from the back of a lorry
Hopes to restart enterprise juggernaut
Windows Live may be hosted from Portakabin-style data centre compounds if Microsoft's 4th generation data centre plans come to fruition.
Microsoft planners envisage thinly-provisioned data centres built from containerised components delivered to site just in time and plugged into a utility spine for power and communications links with standard interfaces.
This fourth generation follows on from generation one - today's data centres which are focussed on uptime, reliability and redundancy. Second-generation ones add in ideas like sustainability, energy efficiency, and the total cost of energy and operations - think hot aisle/cold aisle cooling, among other things. Microsoft suggests its Quincy, Washington data centre (clean hydro-electric power) and San Antonio, Texas facility (cooling with recycled waste water) as examples. The third generation builds data centres taking into account modularity and an even stronger focus on energy efficiency and scale. Microsoft reckons its in-construction Chicago, Illinois data centre with a container hangar and no raised floor will exemplify this generation.
The fourth generation concept is to have all the main data centre elements pre-manufactured and assembled into containers that are shipped or trucked to data centres anywhere in the world. The data centre is a walled and secured compound, not a building, which admittedly leaves it vulnerable to nasty things thrown over the wall. As the demands on its resources grow new modules comprising servers, storage, networking gear or power supplies can be ordered, assembled by suppliers, and shipped to the site for deployment on a just-in-time or thin provisioning basis. They could also be shipped out again when not needed.
It could be like a giant outdoor trailer park. Architecturally it would be as attractive as a a sprawling electricity distribution compound.
Costs should come down 20-40 per cent. Datacentres could be built with varying degrees of redundancy and resiliency as the overall application suite requires. Cooling costs could be reduced by using ambient air instead of chilled air. The use of water for cooling could, the Microsoft engineers hope, be partially or wholly designed out. Site construction costs would be a lot less with significant reductions in time, labour and material costs since no on-site building construction would be needed beyond setting up the ground-level platform or apron, and the power distribution spine.
Suppliers could compete to build the modules with standard interface specifications lowering their costs.
This Microsoft concept is meant primarily, we understand, for data centres facing hyper-growth or needing to be grouped into a worldwide network providing cloud computing resources with edge, anchor and mega-sites. But, if costs became low enough then Microsoft envisages its use spreading.
Find out more information and see a fetching animation video by clicking here. ®
Sponsored: Benefits from the lessons learned in HPC