How does a company measure success in a data centre? This article proposes that there are many key performance indicators data centre managers can use to assess, track, and manage the efficiency of IT facilities and drive business change. In truth, everyone has a different idea about how work is defined in the data centre. Each organisation needs a personalised approach, tailored for their unique business needs, to identify the metrics that lead them to effective data centre management.

Why PUE is not enough by itself
The measurement of a data centre environment is very specific to an organisation. A ratio that indicates something positive for one company might not work for another. For each organisation, it’s about driving performance indicators that truly make a difference from an operational and cost perspective. Only a bespoke strategy can provide illuminating answers due to the specific metrics that result.

In recent years, PUE (Power Usage Effectiveness) has emerged as a popular performance indicator. Promoted by the Green Grid as a metric that can be useful for determining the gross performance of data centre electrical and mechanical systems, PUE shows how much the infrastructure is using versus IT Equipment. IT departments use low PUE numbers to benchmark the impact of the changes made on the infrastructure, whilst marketers cite them as a competitive advantage.


Only well-defined performance indicators provide a true view into the operations of a business.

But is PUE really a standard by which all data centres can be judged? Uptime Institute Executive Director Kenneth G. Brill doesn’t think so.

‘Having measured your PUE, what does it mean, and how do you compare to 1.2 or 1.6, or even a 2.3?’ Brill said. ‘A lower number is better, but where should yours be? Benchmarking is useful only if you are comparing yourself with others with similar power and cooling equipment types, percentages of load utilisation, tier design levels and post codes. Without knowing this additional information, the numbers are virtually meaningless.’ (Kenneth G. Brill, Forbes - Don’t Trust the Marketing Brochures, 2008)

Measuring to business metrics
Marketers may use PUE to create an impression of efficiency, but the bottom line is a conglomerate of metrics, PUE just being one of them, is what really matters. Technologist Mike Manos added, ‘You don’t have to look far to read about the latest company that has broken the new PUE barrier. It’s like the space race. Except that the claims of achieving those milestones are never really backed up with real data to prove or disprove it.’ (Mike Manos, Loose Bolts Blog, 2009)


A company’s data centre metrics should look at the individual components of operations and actually drive business change.

Many data centre managers like to use PUE because it gives them a number that’s always going lower. It looks good and it’s easy to understand. But it doesn’t tell the whole story. A much more direct answer to whether the data centre is a good investment is revealed in detailed performance indicators. Software management tools, for example, provide an inside view of network bottlenecks, application performance, adherence to service level agreements (SLAs) and storage capacity. Cost performance metrics must also be a key component of managing the data centre. This means correlating power usage—the main operational cost—with applications, transactions and SLAs.

A data centre measurement approach should emphasise business metrics—performance indicators that look at the individual components of operations and actually drive business change for the organisation. A financial institution may manage to the cost of transaction processing. Are tradings costly or not costly? Are online banking sessions efficient? The institution has to set up a specific IT performance indicator to get those answers. A manufacturing company has an entirely different set of questions they need answered. For example, they need to know the cost of their supply chain IT and asset management. They have no financial transactions to worry about, so their approach to developing metrics is obviously different. Even Google, whilst widely touting their PUE performance, manages to business metrics by measuring profitability for every ad served.

Accomplishing data centre efficiency
The push for data centre efficiency is greater than ever. Three steps to take when establishing your approach include:

• Drive the metrics that truly make a difference - the best way to achieve a more efficient data centre operation is to identify the most ideal performance indicators for a business and the cost aspects of a business. The metric should be about the system and how it’s performing in relation to the reason it’s there.
• Adopt a rigorous measurement process that meets your organisation’s needs - Measurements on the enterprise management side need to be done in a meticulous and consistent manner. There are software tools that help accomplish these tasks. On the power side, IT people jot down numbers on their clipboard.
• Close the gap between business and IT needs - Executives and business managers must work with IT to make intelligent technology decisions. Factor in the three-year cost of power into the purchasing cost analysis for equipment acquisitions.

Businesses will continue to use PUE in the game of ‘my data centre is better than yours,’ but establishing performance metrics that truly measure success in the data centre is about asking the right questions and searching in the right places. Do that, and illuminating answers to unique business challenges will appear within key performance metrics unearthed from your data centre.

The author: Joe Polastre is CTO at Senitilla.
The views expressed are those of the author.

Related News: Where next for data centre metrics?
Related Video: Standardising Energy Efficiency Metrics on PUE
Related News: Green Grid signs data center efficiency pact with BCS