Mike Manos is extremely busy now and doesn't blog as much as he used to, but when he does post he still gets lots of traffic. Mike and I were laughing once when a corporate data center blog was discussing proudly how many people they reached in a month with their blog. Mike said, "I get that much in less than a week." Which brings up a good point of how your view changes if you knew what you don't know. If they had known Mike gets 10x more traffic than them in a month, they'd wonder how influential they really were, and whether they are successful or not.
I always enjoy reading Mike's post, figuring out ways to use less words than he does, but also reading what Mike is trying to say, but hasn't put it down in words. Luckily leveraging multiple discussions with Mike, I can take a pretty stab at what he was thinking of when he was writing.
Mike posted July 14 on Site Selection.
Site Selection,Data Center Clustering and their Interaction
July 14, 2010 by mmanos
I have written many times on the importance of the site selection for data centers and its growing importance when one considers the regulatory and legislative efforts underway globally. Those who make their living in this space know that this is going to have a significant impact on the future landscape of these electronic bit factories. The on-going long term operational costs over the life of the facility, their use of natural resources (such as power) and what they house and protect (PII data or Personally Identifiable Information) are even now significantly impacting this process for many large global firms, and is making its way into the real estate community. This is requiring a series of crash courses in information security, power regulation and rate structures, and other complex issues for many in the Real Estate community.
What I think Mike is trying to say is it is much easier to build a better performing low cost data center with the right site. But, few understand the complex relationships that affect data center performance. I've always found it naive and over simplistic when companies and consultants use a long list of weighted criteria as the method to pick a site, assuming the highest score is the best site. This works for those who the most complex math they are comfortable with is multiplication and addition, but think about this hundreds of millions of dollars of CAPex, OPex, and IT equipment will be spent over a data center lifetime, and you are going to make the decision based on addition and multiplication?
I believe modeling techniques should be applied to ask the question "what is the right site?" And, Mike has his own mental models of what is right and wrong.
The right site for what? Pick 3 - 5 data center designs that you think you would want to build and use them as models to represent what you intend to build. If you have built the model with enough detail you should see the relationships that Mike talks about.
Tying into the power conversation is that of water. With the significant drive for economization (whether water based or air-based) water continues to be a factor. What many people don’t understand is that in many markets the discharge water is clean to dump into the sewage system and to ‘dirty’ to discharge to retention ponds. This causes all kinds of potential issues and understanding the underlying water landscape is important. The size of the metropolitan sewage environments, ability to dig your own well efforts, the local water table and aquifer issues, your intended load and resulting water requirements, how the local county, muncipality, or region views discharge in general and which chemicals and in what quantities is important to think about today. However, as the use of water increases in terms of its potential environmental scrutiny – water is quickly rising on the site selection radar of many operators and those with long term holds.
I have friends who designed a waterless cooling system in Australia due to the drought conditions. There was a cost associated for this data center design vs. cooling towers, but when you looked at the total system it was the right design.
If you really want to be advanced you can use semantic models.
Savanna is a model-driven analysis solution for solving complex problems. The magic of Savanna is in defining models that address what’s relevant to your problem at any given point in the analysis process. Savanna’s semantic models are driven by the Thetus Publisher architecture, enabling information synthesis by offering users the unique ability to derive meaning from information sets and to bridge the gaps in information. Savanna’s innovative, mind-mapping interface provides intuitive tools for approaching analysis from a point that frames the problem rather than one that starts from the information out.
Warning this technique works, but few have the capabilities to operate in this way. Using semantic models change the game as you focus on the problems and questions to ask, and enables you to see things others cannot.
To take control of site selection you need to have data center designs in mind for what you are building. If you are Mike Manos you can see the relationships of the site to the data center designs and how the system will operate.
If you can't do what Mike Manos does, then be prepared to make lots of mistakes even if you hire experts. Because you are not in control and are being told what to do. Do you think you can be a good cook by hiring a bunch of experts to tell you what to do? You need to be in control. Use data center designs to take control.
If you walk into a site selection consultant and say here are five data center designs I am looking at find me sites that support these designs. When you find me a site tell me which design works best and worst on the site. You'll find out whether the consultant can do more than addition and multiplication, and whether they really understand what a good data center site is.