Data Center

Data science aids compliance with green data center standards

When people in business hear the term “science” they often think of pure science—PhDs in lab coats doing fine research but not finding a practical purpose for it. For those who have been following my recent posts on the emergence of data science for data centers, you can be rest assured—this type of science is far from impractical.

A case in point for the practical use of data science for data centers is in helping companies comply with sustainability standards. In Asia Pacific, a prime application is in meeting Singapore’s SS 564 Green Data Centre Standard. The aim of this standard is establish a continuous improvement framework to improve the energy efficiency of data centers in Singapore, a country which needs to import most of its energy. There are other green data center programs for which data science could help bring efficiencies, such as NABERS in Australia.

The SS 564 program leverages metrics as part of a “Plan-Do-Check-Act (PDCA)” improvement methodology. Here is where data science shines—in spotting trends and nuances that when acted upon, move SS 564 metrics in the right direction. It’s worth noting that SS 564 doesn’t dictate rigid thresholds for everyone in the program—it’s more about establishing continuous improvement and then reporting and checking up on progress for each organization.

Let’s break down how data science can help with SS 564 in more detail by considering the metrics involved, which include:

  • Power usage effectiveness (PUE)
  • Energy distribution factors
  • Ambient relative humidity
  • Lighting density
  • Carbon usage effectiveness (CUE)

Several other metrics also a typically tracked, such as air flow efficiency, return temperature index, uninterruptible power supply (UPS) load factor, server equipment load density, and cooling system efficiency.

That’s a lot of metrics, right? Even if a company operating a data center had help from a provider such as Schneider Electric in establishing SS 564 metrics, and in deploying data center infrastructure management (DCIM) software to dynamically track those metrics, the reality is that in many cases, the user company may not have the expertise or the time to apply the tools and manage the metrics as part of a PDCA methodology. And, even if they had good people skills for monitoring a dashboard, it’s also quite possible the correlations between trends would be very difficult to spot just by watching the gauges.

Here’s where data science helps by uncovering and/or predicting hard-to-spot trends. Data science extracts useful information from data sets, and unlike traditional business intelligence, it uses modeling and other analytical techniques to deliver predictive insights.

Let’s say, for instance, that PUE begins to degrade. It may be that the root cause is tied to increasing use of server virtualization, and rapid shifting of virtualized loads without scaling back of the power protection and cooling capacity quickly or precisely enough.

As a result of these complexities, PUE suffers. While this dynamic seems simple enough, it becomes complex to solve by monitoring alone for bigger, heavily virtualized data centers or multiple data centers that share IT loads. What’s needed is some predictive analytics—data science—that pinpoints outcomes and adjustments. The same holds true for other dynamics, such as how rapidly changing weather impacts cooling system efficiency.

In essence, data science is great for guiding decisions involving larger sets of dynamically shifting data. It’s akin to the way that digitized systems in automobiles have evolved to the point where we not only have reliable anti-lock braking, but also electronic stability control systems and even collision avoidance systems.

I’m not suggesting that data centers seeking to comply with SS 564 or other green data center programs can comply simply through data science. There is no “auto pilot” for SS 564 compliance, but data science can serve a supporting tool for understanding rapidly changing conditions and predicting the best response.

In reality, companies looking to comply with SS 564 typically also may need baselining and assessment services, help with DCIM and metric/dashboard creation, as well as help with upgrades that simulations show will make operations significantly greener, such as air containment, or perhaps new equipment with economizer modes of operation. When you combine these sort of services with data science, compliance with standards such as SS 564, or with internal sustainability efforts, become more effective.


No Responses

Leave a Reply

  • (will not be published)