Data center engineers looking to maximize energy efficiency can now model design changes using a tool called Prognose from U.K. startup Romonet.
The software lets engineers construct a graphical model of their data center, including the power and cooling equipment and how it's all laid out, and then make changes to the model and see what impact those changes would have on overall energy consumption and efficiency.
The idea is to help companies find the optimal design for a data center, and reduce the risk of making a poor decision about where to locate air conditioning units, transformers and uninterruptible power supplies, for example, that could cost millions in wasted electricity.
Prognose was launched in the U.K. last November, and Romonet plans to announce at the DatacenterDynamics conference in San Francisco this week that the software is now being sold and supported in the U.S.
It's aimed at data center design companies like Skanska USA and Syska Hennessy Group, which Romonet said have both been trained on the software, and also at large enterprises with their own data center design team.
Jvan McGuire, a vice president with Syska Hennessy, said his company has just started using the software at two customer sites, one a new build and the other a retrofit. It provides a way for it to calculate PUEs accurately, he said.
PUE, or Power Usage Effectiveness, is a data center efficiency metric. It can be measured in a variety of ways, and results can vary widely based on factors such as climate and utilization levels. Prognose lets Syska Hennessy calculate PUE scores in several ways and present customers with the results they find most useful, McGuire said.
"Everyone's very concerned these days not just with the reliability of their data center but also the energy profile, the carbon footprint," he said. "This is a very good predictive model that helps us take a hard look at each piece of that."
Prognose can also provide its output in dollars, calculating the operational cost of different design choices, said Romonet CEO and co-founder Zahl Limbuwala. "You end up with something you can take to the finance guy and say, 'This is why we want to invest this money, because these are the savings it will result in,'" he said.
One limitation is that Prognose doesn't monitor the actual energy usage of equipment. Instead, Romonet supplies a database of information about various products and configurations that customers use to populate their model. They can also enter data about climate, utilization levels and other factors.
Beyond that, they will need to purchase monitoring equipment separately if they want to incorporate real-world data from their equipment. McGuire said Syska Hennessy hasn't used Prognose for long enough to say whether monitoring equipment will always be a requirement, but he said most of his customers would "probably want some kind of monitoring regimen to go along with it."
Limbuwala said that may not be necessary, however. "One of the interesting things we've found is that the performance of individual devices is relatively unimportant to overall data center performance. The topology of the data center -- what powers what and what's cooled by what -- and the control scheme for those devices is much more important," according to Limbuwala.
Customers can often get "vastly better performance" from their data centers with existing equipment just by controlling it more intelligently, he said.
Romonet sells a version of Prognose for consultants, on a per-user subscription basis, and a version for enterprises that carries a license fee based on the size of the data center in megawatts. Romonet declined to provide actual prices for the software.
Managing energy use in data centers is a hot market right now and dozens of startups have sprung up in the space. Rising energy costs, capacity constraints and the threat of carbon legislation (or the reality of it, in Europe) all have forced companies to take a closer look at energy usage.
"We rarely talk to anyone these days who is not interested in the energy performance of their data center," McGuire said.
Limbuwala founded Romonet five years ago with CTO Liam Newcombe. Both had worked at telcos and managed service providers, where they said they were frustrated at how difficult it was to calculate the actual cost of delivering a data center service.
They claim their predictive modelling approach is unique in the world of data center infrastructure management, or DCIM, and are trying to coin a new acronym for what they do: DCPM, or data center predictive modelling.
The company was in "stealth mode" for a long time, he acknowledged, and writing the software was a big undertaking. "It was almost a boil-the-ocean approach in what we were trying to solve," Limbuwala said.