In the data center world the technologists and engineers who design, operate, and utilize data centers can frequently be at odds with the real estate money guys and the marketers, who have more interest in accelerating sales, revenue, and investment returns.
Claims to impress investors and customers are frequently not presented in the way the operations pros and the technologists who run the data centers would prefer. As in all fields, that includes using numbers to make your case, and includes the misuse of some well-intentioned metrics.
The claim game is partially to blame for some changes with the measurements that have long been part of the data center conversation. For years, major industry metrics included PUE (power usage effectiveness) to measure how efficiently a data center uses energy, and Tier rating, which measures resiliency in regards to uptime and availability. But over the years both of these metrics were misused by marketers and others with an agenda trying to make an impression.
PUE: PUE will not be a part of the AHRAE data center standard, which is used by many local governments in their permitting and building inspection. Instead ASHRAE (the American Society of Heating, Refrigerating, and Air Conditioning Engineers), will move on to a new metric, Design MLC, defined in Equation 6.2.1, 1: in the current draft, which is available for comment by data center professionals.
When data center operations professionals gathered, you were also likely to hear complaints about PUE claims made by others. Numbers as close to 1.00 (the lowest possible PUE) as possible were frequently used by marketers and developers and to engineers who knew the facilities, those calculations didn’t always pass the sniff test. “I’ve earned the industry’s first 0.98 PUE” announced one data center facilities pro from the podium of an industry event, to the laughter of his colleagues. One purpose of questionable PUE claims was to attract business from end-users eager to highlight their sustainability achievements, and the easiest path was making a purchase from someone who had apparently done the work required.
The bigger problem with PUE was the fact that it was an unfair metric to judge data centers that were not yet full, as well as multi-tenant data centers whose colocation customers didn’t use their space optimally, didn’t use the most efficient gear, and failed to commit to thermal and airflow best practices within those leased environments.
Misuse of statistics has led to the demise of widely-used measurements in many industries, and we may be seeing the same with PUE. PUE claims were often a source of controversy, and may have contributed to souring people on the use of the metric. But as with with any new metric , it is likely only a matter of time before we see mini-controversies develop over the validity and importance of certain Design MLC claims.
Tier Rating System: Marketers and data center developers deserve more blame for the decline and fall of the tier rating system, which was junked by Uptime Institute last fall.
One persistent issue was the misuse of tier ratings. Providers engaged in self-rating, calling themselves Tier 3 or even Tier 4 without going through the thorough and expensive process of evaluation by Uptime Institute. Uptime’s use of roman numerals for Tier I through Tier IV led many to claim things like “Tier 3 design” in their materials. A salesman leading a data center tour was unlikely to go out of his way to delineate the difference to his customers. And why would a provider invest in a tier rating when their competitor across town was making the same claim without the same investment and process?
Could the misuse have been avoided via improved branding? Had Uptime branded the Tier system with something more unique than a simple number, they might still be collecting significant fees for ascertaining the reliability and availability of data centers through the tier system. In contrast, the US Green Building Council (USGBC), has branded their certification quite well. LEED (Leadership in Energy and Environmental Design) is more difficult to claim or mimic. A number (Tier 3 vs. Tier III) is easy to knock off. LEED Gold less so.
The expense of earning a tier rating from Uptime was a factor referenced by providers. But that’s the nature of the certification process and achieving LEED is burdensome and pricey as well. Many providers go through the process to obtain LEED and conversations about false LEED claims are not common.
Customers Share Blame: Some knowledgeable customers certainly downgraded some data center options based on self-tiering, but too many claims went unchallenged. Not enough customers pushed back by immediately rejecting the self-rated tier claims of some providers and developers. If they had, it isn’t a stretch to think that the Uptime Institute tier rating might still be around.