Is the modular datacenter more efficient than traditional raised floor?

The Arizona Public Services electricity utility takes a look at IO datacenters.
Written by David Chernicoff, Contributor

IO datacenters released a study this week, performed by third-party Arizona Public Service that shows that in their Arizona facility, the IO modular technology delivers better PUE than the traditional server room model in the same facility. Using OPUE as the comparison point, the open room had a rating of 1.73, while the closed modules had a rating of 1.41, which IO claims could result in energy costs savings of over $200K per year.

One this I’ve always liked about IO is that they’ve instrumented everything that they can think of in their datacenter facilities. This gives them a really detailed view of what is going on, its impact on operations, and with analysis, what can be done to improve future performance. It’s a step beyond add-on DCIM technologies, but, of course, applies only to their own facilities at this level of detail.

The APS (Arizona’s largest public utility company) study was able to take advantage of the information the facility can provide and was run against an operational, IT loaded , facility, not a projection of what a future or planned facility would be able to do. They are able to get value from the comparison because both types of operations are run in the same facility:

  • Same Geography -- All numbers from this report come out IO Phoenix, which contains traditional 1.0 and modular 2.0 data centers.
  • Same Chiller Plant -- Shared chiller plant – same efficiency for both environments.
  • Same Building, Same Envelope --  The IO.Anywhere modules are housed in a data center
  • Same Operator --  IO operations staff monitor and maintain both environments on behalf of customers and IO.
  • Utilization -- Current utilization of a 200 kW IO.Anywhere data module was used to project annual savings based on the historical data of the operational facility

I’m also not surprised at the results. Controlling environmental factors that directly impact operational costs in a datacenter is much easier to do in a small space than a large one. Using the same chillers and applying the technology to a tightly controlled module is always likely to have better efficiency than is a large, raised floor datacenter. So to that end, the results of the report vindicate the IO plans to move towards the modular architecture for their facilities. Their plan to pass the savings on to cust9omers could result in a competitive advantage for businesses willing to take this approach with their colo and hosted solutions.


Editorial standards