Can Velocity Revolutionize The Data Center Business?

James Staten


Digital Realty Trust is trying to do to the data center business what Dell did to computer manufacturing and Wal-Mart continues to do to the retail market — make delivery efficiency the killer app. It claims it can build out a complete enterprise-class, energy-efficient data center in as little as 26 weeks (versus 16-24 months by its competitors) and for less than half the competitive cost.


Bringing velocity and low price to data center construction is certainly a new concept and one that could have revolutionary implications. This area of design and construction has been very specialized, with each project being unique and the magnitude of a single order in the $50 million range. DRT, a REIT focused purely on the data center market, has made a name for itself by building and operating data centers for corporate clients. As a REIT, DRT has held onto the facility, leasing it back to the client or using it as a collocation facility.


But this week it announced its POD Architecture Services through which it will build you the data center of your choice and you can own it at the end of the process.


How does DRT deliver against these dramatic claims? By borrowing from Henry Ford, Dell, and Wal-Mart and bringing the best of their concepts to this staid business:

  • First, DRT borrows from Ford by offering a standardized product its POD Architecture (the proverbial Model T available in black or… black). This modular data center is a self-contained set of computer room rows outfitted with highly energy efficient and self-contained power and cooling facilities. A POD isn’t a data center but a fixed segment of a data center. The approach allows you to build out only the portion of the data center that you plan to occupy in a given timeframe. This way you don’t carry the full burden of the entire data center build out in year one, if you plan to grow into the facility over the next 10 years. The design also lets you to mix and match availability tiers within the same facility, a concept HP’s EYP also favors.

    Image: http://www.digitalrealtytrust.com/images/dlr_pod.jpg


  • Second, DRT borrows from Dell who have mastered manufacturing and inventory turns just in time. In the mid 1990s Dell perfected its assembly process by having its suppliers place warehouses adjacent to its assembly plants and hold the inventory until just minutes before Dell needed them to fulfill already-placed orders. This allowed the company to achieve 90+ inventory turns per year and maximize the profitable delivery of low margin, high quality products. DRT employs similar thinking. Since it builds nearly every data center as a collection of PODs it is able to make bulk orders for long lead time items such as diesel generators, PDUs, and economizers and thus have them available just in time for each construction project. Long lead items alone can put a data center project behind by a year or more. DRT, due to volume, goes right to the front of the line for its suppliers.

  • Third, DRT leverages its buying power to drive down the costs and thus prices of the components it assembles by sending the specifications for its PODs and all future enhancements to current and potential suppliers who, because of the volume DRT can drive, will build products to suit their requirements. This approach could lead to higher cost and proprietary components but DRT’s aim is to drive down costs and drive up repeatability, so their specifications are open and interoperable. Most of the enhancements DRT requests end up in the suppliers mainstream products.

Combine these tactics together, open up your market to all takers and you have a recipe for even more efficient data center delivery. DRT believes that through increased supply chain velocity it can drive data center build-times down to as little as 16 weeks and wring out potentially 20-30% greater cost savings. If so, others in the market will be hard pressed not to follow their lead.


And smart enterprises will be remiss not to invite DRT to bid on their next data center.



By James Staten


Check out James' research