Posted by Todd Boucher
on 05/18/2011
If you attended the Uptime Institute Symposium in Santa Clara, CA last week, you may have left wondering what the term ‘modular data center’ really means. The idea of modularity was hailed during countless presentations as a requirement for any organization looking to build a new data center and the best way to ensure energy efficiency. Unfortunately, none of these presentations defined what exactly a ‘modular data center’ is and instead used the term to describe a number of different applications. I heard a ‘modular data center’ described as:
- A data center “POD” - prefabricated and shipped to your site ready to deploy
- A methodology whereby you create “building blocks,” or pre-defined increments of growth. For example, if you have 5,000 square feet of data center space with 500kW of IT load capacity, you could build 5 (five) individual 1,000 square foot, 100kW modules at a time.
- A container holding infrastructure components like chillers, switchgear, and UPS systems that you “plug in.”
- A framework for a cloud-ready data center that integrates processes and personnel to rapidly scale to meet growing XaaS demands.
For an owner, creating a modular data center could involve one or more… or all… of the above. What is lost in describing all these various forms of modularity is how to help owners understand what is best for them. Organizations like Facebook, Google, and Vantage that presented at Uptime are in the data center business, or at minimum have the internal expertise to articulate their specific data center requirements. I would argue that having this expertise in-house is the exception, not the norm, for most companies.
As a result, it is important for us to realize that before we recommend modular solutions (or any solution) we have to understand the business requirements of the data center and educate the owner on what methodology is best suited to satisfy those requirements. Terms we used to hear all the time -lifecycle, whether you own or lease the data center space, budget, ROI- all still have a significant impact on the determining the best approach for building your data center. For example, if you are building an enterprise data center, own the building, and have less than 1MW of IT load, you probably don’t need a container-based solution as your system isn’t large enough to realize the economies-of-scale. However, if you are a cloud-based SaaS or PaaS provider and want to minimize CapEx investment while having extreme flexibility to grow on-demand, you will likely want to a consider POD or container-based approach.
Vendors deserve a lot of credit for their continued innovation and for the unique solutions on showcased at Uptime. An owner can and should consider all these, but before doing so should take the time to define the business objectives of their data center and use that knowledge as a roadmap in evaluating potential solutions. Until every organization we work with is the exact same, we can’t and should not treat the data center like a commodity.
