Notes from the Consultant’s Jungle

When launching a new data center build project, where the data center will be located is a fundamental issue. There are many factors in deciding where the data center will be, but all of these factors can arguably be consolidated into two issues- Risk and Cost.

We mention risk in terms of Risk Management. Even if a data center is not specifically a disaster recovery site, many issues involving the physical location of the data center are evaluated to assess risk to availability of the equipment and data that would reside there. For example, exposure to environmental threats such as flooding, storms, earthquakes, and so on is often evaluated. Man-made environmental threats such as proximity to chemical plants, railways, gas lines, and so on are included here too. Risk management evaluations will also consider factors such as local crime rate, political stability, and threats from war or terrorism. For a security and risk management professional, this list is long, but any risk exposure is also prioritized and weighted for pragmatic consideration when evaluating site selection for a data center.

In terms of cost, there are two major components to consider. The first is the MOOSE cost. MOOSE is a term coined by Forrester to represent Maintenance, Operation, and Ongoing Support Expenses. It essentially means the costs to “keep the lights on,” as they say. Support staff, maintenance, and utility expenses are big line items in the MOOSE budget.

When we talk about MOOSE costs, a lot of attention falls on the subject of energy costs, since the utility costs of data centers is such a large portion of the operational expense budget, and draws attention to data centers as high consumer industrial facilities. This is a very important point.

These points are brought together in a recent article in Information Management by Mel Duval quotes a report from Tishman Technologies that suggests a “Top Ten” list of best countries in which to build a data center. This survey proposes the best countries in which to build a data center, largely based on the risk elements mentioned above. Here are their conclusions:

1. Iceland, Reykjavík - Low-energy cost, free cooling

2. United States, North Carolina - Low-energy cost, favorable labor, and fiber optics

3. China/Vietnam; Shanghai -Extraordinary demand and new/diverse fiber optics

4. Latvia (no city given) - Low-cost hydro power, favorable labor

5. India, Mumbai - Extraordinary demand, favorable labor

6. Russia, St. Petersburg - Favorable telecom, free cooling, favorable labor

7. Canada, Vancouver - Favorable labor, free cooling, favorable telecom

8. Japan, Toyko - Extraordinary demand, favorable telecom

9. New Zealand (no city given) -Strategic location, green power, favorable labor

10. United Arab Emirates, Dubai - Strategic location, favorable labor, telecom

This is an interesting list, but I’d like to contrast with a bit of pragmatism (that no doubt the authors of the list are aware of).

The idea of identifying countries for locations of data centers based on risk and access to low cost energy is an academic exercise that not many enterprises can take advantage of. Aside from the Web 2.0 and mega cloud providers, not many enterprises have the ability to so broadly cast their site selection net.

I’ll choose Iceland to make my point. Iceland is a good pick for the number one spot as defined by the study. It comes out on top of studies like this because of natural geothermal power sources, lots of free cooling, and it’s a stable political environment. From that perspective, it’s a good choice.

The problem though is all the other difficulties around placing an IT footprint there, if you’re not already stationed in that country. It’s one thing for, say a New York-based company, to put a data center in North Carolina and recruit the support and operations staff to run the facility. Can you imagine trying to do that for a site in Iceland? Who is going to build the facility and project manage the job? It’s hard enough to manage such an undertaking in your own country. So my point is that if you don’t already have a presence in the country of interest then the benefits of lower energy costs through operating there are going to be pushed to the side.

There’s an important point about global IT infrastructure that is often overlooked in discussions about data center locations. Increasingly since 2001, there are legislative and regulatory constraints around electronic information storage and transmittal. In North America, where I write this article, two of the most obvious in this regard are Homeland Security and PIPEDA. If you have customer data stored and processed in your data processing systems, it is a complicated issue to transmit and store that information across political borders.

In the US and around the world now there are emerging legislative factors that specifically impact facilities that are large consumers of energy (i.e., data centers). This legislation is focused on carbon reduction, in many cases by levying tax penalties on heavy consumers of energy. Bills such as Waxman-Markey now coming into the light of day are going to cause significant financial pain for those enterprises that recently invested in large data center construction projects.

The other issue of course is the whole idea of building a data center. Even in more favorable economic times, only the largest of enterprises can cost-justify investment in creating and maintaining a purpose-built data center. The much more attractive option is colocation. With colocation, you avoid the capital investment (huge, by the way) of building mission critical infrastructure that data centers need. You also avoid the burden of creating your own world-class operations staff and procedures that are the real drivers of high availability. In other words, you concentrate on what your business is good at, and leave the data center expertise to the data center experts.

When in a colocation model, the issues around energy costs shift perspective dramatically. Sure, you may want to demonstrate “Greenness” in some way and have an expectation that your colocation provider is Green-aware. However, you’re leveraging a shared power/cooling infrastructure with the other tenants in the colocation facility so the Green topic shifts to those energy management options above the white space, which you do control directly and are much easier to demonstrate energy efficiency results.

This brings us back to the question of where the data center should be located. By choosing a colocation provider with presence in a large number of active business markets can yield benefits to the enterprise. Chances are that the provider will have a site very close to where your IT staff is already located. Chances are too that the provider’s sites in other markets can yield favorable options for BC/DR planning. Also, a colocation provider with national scope can be leveraged for excellent network density and carrier availability. Finally, if your customers are also located in that provider’s facilities it’s possible to take advantage of significant telecom cost savings by connecting to customers within the already-built shared infrastructure.