Telcos Need Data Center Education
In fact, the metric most often cited as representative of the overall "green" health of a data center, the Power Usage Effectiveness (PUE) metric, doesn't provide the entire picture.
During recent months, leading-edge data center operators eBay Inc. and Facebook have announced energy efficiency dashboards that show how much energy the companies' facilities are consuming -- in real time, in the case of Facebook -- and they're trying to get other companies, telcos included, to adopt their methodologies.
Instead of just looking at PUE, they incorporate additional criteria, such as overall carbon output and the amount of water used for cooling, to get a better understanding of data center performance.
Facebook announced recently that it is publishing real-time energy consumption data for its data centers in Oregon and North Carolina, while eBay announced its Digital Service Efficiency (DSE) methodology and said it will publish efficiency results quarterly.
Google started the trend towards publishing efficiency data back in 2008, but so far the company has not gone as far as Facebook or eBay in delivering an online dashboard.
For their part, telcos have operated their own large internal data centers for years and, increasingly, their cloud and content delivery services require them to operate external data centers as well. Whether they will embrace the eBay and Facebook methodologies is unclear. Sprint Nextel Corp., for one, says it won't anytime soon.
"We don't use DSE and believe it's unlikely we would in the foreseeable future," Sprint's VP of IT, Josh Morton, writes in an email to Light Reading, adding that Sprint will continue to focus on absolute reductions and PUE for now. "We believe it may be much easier for eBay to use this measure as it is a good match for its transaction-based, operating model."
The PUE metric was originally developed by the Green Grid, a non-profit industry consortium created to improve energy efficiency in the ICT industry, and is a ratio of IT equipment versus the overall energy consumed in a data center facility, which includes power for things such as cooling, power distribution and lighting.
There are a couple of problems with relying only on PUE to determine data center energy efficiency. First of all, a company can have a perfect PUE score and still be getting 100 percent of its energy from coal, which isn't terribly green. Second, as IT facilities become more energy efficient, it's actually possible for the PUE number to go up.
"That's the fatal flaw of PUE," says Tom Raftery, principal analyst at GreenMonk, the Energy and Sustainability practice of research firm RedMonk. "It's designed to measure the efficiency of a facility, not the IT equipment."
New measures of greenness
That's why eBay developed its DSE metric, which it describes as a sort of "miles per gallon" measure for technical infrastructure. The DSE tells eBay exactly how many business transactions -- in this case, buying or selling -- are completed per kilowatt hour (kWh) of energy consumed, the cost per transaction, the environmental impact in metric tonnes of carbon dioxide per kWh, and the revenue generated per kWh.
Similarly, Facebook's new real-time efficiency dashboard looks beyond the PUE and measures water usage effectiveness and temperature and humidity at its data center sites. The company is making the front end for its dashboards available as open-source code so that any organization interested in sharing efficiency metrics can use the dashboards to get started.
Telcos can learn a lot from companies such as Facebook, eBay and Google. Even if they can't afford to build state-of-the-art data centers powered by renewable energy, they can take steps to figure out how much energy their existing data centers are consuming and make them more efficient.
The first step is measuring PUE, carbon output and water usage. As Jeremy Rodriguez, distinguished engineer, Global Foundation Services, at eBay explains, "If you can't measure those, it will be a challenge to go after a DSE-type metric. But even if the numbers are rough and fairly inaccurate, it's better to look at them than not."
-- Dawn Bushaus, special to Light Reading