Want to build your new data center faster, better and cheaper? Try containerized, modular data centers, says The Green Grid (TGG), a group dedicated to advancing resource efficiency in data centers and business computing systems.
The Green Grid, which introduced its Carbon Usage Effectiveness (CUE) metric last year, just released a report on using containerized and modular data centers (you can download the full report here). The report, authored and edited by member companies of The Green Grid, points to the increased use of containerized and modular systems and how this trend is resulting in a faster deployment, lower operating and capital costs, and the potential for higher density and energy savings. The report also delves into the various aspects an organization should consider during the planning, design, deployment, operations, and decommissioning phases of their data center facility lifecycle.
I’ve covered modular data centers in this blog already. You can read here about Orion, a New Zealand utility whose data center was damaged by the deadly earthquake in that country last February, a magnitude 6.3 quake that left 181 people dead and billions of dollars in damage. The company turned to a modular data center system to replace what it lost.
These modular systems are typically data centers packaged together in portable, standard forms and include server racks, power and cooling units, redundant power supply hooks and even networking gear. Generally speaking, a modular data center can be fully functional yet built in a pod-like form, transported to any location and set up in weeks or even days, and can be built to meet the specific requirements and support multiple technology vendors and multiple systems in an industry standard rack environment. Should additional capacity be required in the future, more of the pod-like containers can be added.
TGG’s report adds that a containerized or modular data center facility separates some or all of the core facility areas of a traditional data center into discrete, prefabricated components that are assembled at the target location, and not that the modular center is a “set of one or more prefabricated and pre-engineered metal buildings (PMBs), along with any optional and required building infrastructure equipment (which may or may not be PMBs).”
There’s plenty more detail in the report that further delineate the differences between traditional data centers and containerized or modular ones. But the information I think is particularly interesting is the section discussing ownership and contracts. I say this because these new types of facilities will require organizations to rethink the ways they write contracts and craft ownership models.
For example, the report points out that a traditional data center with a raised floor may be depreciated over 15 or 20 years, while a containerized facility may depreciate on the same schedule as the servers it houses – which, of course, is a vastly different depreciation model. The report goes on to describe three different containerized or modular ownership models. There’s the conventional model, in which an organization owns both the containerized facility and the IT equipment within it. In another model, the organization could lease the containerized from the manufacturer, (a decision that might make sense for an organization that expects to add servers over time rather than deploying a fully populated facility, the report notes). The third model described is one in which an organization’s IT resources are fully outsourced or virtualized to cloud and hosting providers on an SLA basis, and those hosting providers use containerized facilities. In this case, according to the report, the containerized facility is “abstracted from the customer organization, just as a physical server is abstracted from the end user in a virtualized environment.”
There’s plenty more information in the report, and I think it’s a good starting point for companies that may be considering containerized or modular data centers. Take a look.