The Uphill Struggle For Data Centre Efficiency
After years of effort, why has the efficient data centre movement made so little impact, asks Peter Judge
Data centre efficiency has been a major cause in the IT industry for some time, but it sometimes seems to have made almost no difference.
When we launched this site three years ago, efficiency was a hot topic. the worst of the recession was still ahead, but it was very clear that energy costs would be going up, and money would be tight. Cutting the energy used by data centres was a double winner, saving the environment by reducing greenhouse emissions, and saving money too.
Has efficiency lost its lustre?
Since then, we have heard of a steady stream of new data centres, all of which have very low PUE (Power Usage Effectiveness – a low score means less energy wasted).
At the same time, Greenpeace ran a fierce campaign against energy-guzzling data centres, focused on Facebook’s Prineville location, uses electricity generated by coal-fired power stations. The two companies buried the hatchet over Christmas, and Facebook’s new data centre in Sweden (shown here) will be very energy efficient.
Meanwhile, we have also followed the saga of the UK government’s CRC Energy Efficiency Scheme, which started as an ambitious and complex cap-and-trade scheme that would have imposed a very complex system of carbon accounting and rewarded those who played the game best with real money.
Instead, CRC is now a straight green tax, with a “league table” bolted on, which allows some firms to boast about how green they are. It is a league table in which, incidentally, IT firms perform very badly.
After all that, it is a big surprise to find that the UK’s data centres are not doing anything to cut their energy emissions. Admittedly, the fugures come from an online survey with only 70 responses , but the finding that three quarters of those data centre owners expect to increase their energy use is a shock.
It’s even more of a surprise to find that amongst those who understand PUE, figures of more than 2.0 are common – and around a third of the sample didn’t even know what PUE is. The Power Usage Effectiveness measure was supposed to be simple and easy to promote, so it’s a big disappointment to see it ignored.
Of course, a bugger and more rigorous survey might show something different. In particular, it would be good to see a larger number of people responding, and weight their answers according to how much data centre space they actually have.
I would bet that the ones who are ignorant of PUE are the small players, while people like Facebook, Apple, Amazon and Google, who use megaWatts of electricity cannot possibly be ignorant of efficiency benefits.
As the cloud aggregates more data centre processing in larger sites, I would expect a larger survey would show overall efficiency rising – however the overall total energy used will also be going up, since the cloud is driving greater and greater energy usage.
In other words, energy efficiency is just one of many factors in play here. I would hope that efficiency is playing its part, whatever a small survey may show.