|
In today’s environment, any service provider worth its salt would be invoking concepts such as life cycle management or total cost of ownership, but right now I am increasingly concerned about pretty fundamental stuff.
 Bottom line, the average data center is becoming environmentally obsolete at an alarming rate.
With the exponential growth of processor capability being combined with technology adoption rates, the resulting performance expectations are outstripping current infrastructure designs and causing real issues.
The manufacturing of transistors on silicon is doubling every 18 months – versus Moore’s Law of Computing’s original prediction of 24 months – and the resulting power consumption is driving the environment around those processors to take a big gulp.
Kenneth Brill, in his insightful research report for Uptime Institute on the subject, recently pointed out that “computational performance increased by a factor of 27 between 2000 and 2006 (but) energy efficiency has gone up by only a factor of eight during the same period.”
Uptime Institute is the premier resource of information for data center managers.
Today, we face a challenge in which our approach to data center infrastructure development may not be keeping pace with our ability to procure and implement advanced systems. In fact, in a 2006 survey of data center operators by the Aperture Research Institute, “Almost 40 percent of respondents reported that they had run out of space, power or cooling capacity without having sufficient notice.”
So isn’t this just a good old planning-quality issue? Can’t we just get better planners, more money or “virtualize” more servers? Wasn’t virtualization going to make all this better?
The fact is that virtualization is actually helping us buy precious time. But virtualization is a one-trick pony, and one that generates a little extra power and cooling consumption of its own.
Now it’s time to think ahead and ask important questions about how we adjust to the rising cost and stability of power, power distribution, cooling, redundancy and general environmental infrastructure. Whether it’s the consistency of the grid, the rapidly rising cost of power distribution or the “double oven” effect of blade servers, density-driven processors or attached peripherals, we need to think ahead and consider our options.
Here are a few questions to ask before expanding a data center:
• Do I have a road map from senior management regarding the critical applications and supporting systems that will become my imperatives?
• Is my current center ready for the change that is occurring, and will it stand up to even more accelerated change? Is my infrastructure positioned to grow at a rate equal to or greater than my processing capacity in the next three years?
• Will the one-time value of virtualization offset any environmental exponential growth?
• Will the necessary power and cooling upgrades and redundancies cause any capital expenditure pressure on my system requirements? Are system requirements also business imperatives that I can’t compromise?
• What controls can I initiate to protect myself from the clear challenge of rising power-distribution costs? Do I have a macro-power stability issue to consider? Do I have the cost of power in my budget?
• What are the alternatives to my current plan? Will my company’s culture allow me to utilize them or should I start the executive education process now?
As a shared service provider, we have seen an unprecedented level of hurry-up defense or unplanned data center expansion in the past 12 months. As in any technology-planning challenge, circumstances change, and we have to be dynamic enough in our ability to re-evaluate and respond.
Ah, progress, such a bewildering mistress.
Manny Quevedo is vice president of corporate development for CoSentry
782-7701, www.cosentry.com
|
