The New York Times has an idiotic “analysis” of data center power efficiency on the front page today no less. The headline: “Power, Pollution and the Internet” should make clear where its going. What starts as a reasonable assertion, that many data centers waste a large amount of energy, quickly degrades into an unbelievable amount of disinformation and poor analysis. And this was the result of “a yearlong examination by the New York Times” which “at the request of The Times, the consulting firm McKinsey & Company analyzed energy use by data centers and found that, on average, they were using only 6 percent to 12 percent of the electricity powering their servers to perform computations“.
To this point, that may be true, but also note that they don’t mention anything about the data other than “the study sampled about 20,000 servers in about 70 large data centers spanning the commercial gamut: drug companies, military contractors, banks, media companies and government agencies.” The first warning sign of where the article goes bad is “Most data centers, by design, consume vast amounts of energy in an incongruously wasteful manner, interviews and documents show.” Yep, all the customers I work with purposely set out to waste as much energy as possible. Uh huh… Many do, but not because they want to.
Where it really goes off the rails completely is that the remainder of the “analysis” goes After Amazon in particular and “cloud” datacenters. From both the industries they cite as being in the sample and by the numbers they found, they obviously did not look at industry leaders, or weight by server count, or any number of other things which might have resulted in a proper analysis. The point they seem to being trying really hard to drive which is that many datacenters are inefficient is absolutely true, but the “gotcha” journalism and wanting to go after an example they think everyone will recognize “cloud” and “Amazon” they lose the main point which is that it’s regular companies, i.e. their “drug companies, military contractors, banks, media companies and government agencies” which are the ones with crappy datacenters in many cases. They are why Microsoft, Amazon, and Google datacenters exist and are highly efficient! This inefficiency is why cloud computing is growing so quickly: to provide an alternative to individual companies having to maintain their own datacenters. What they are talking about, without mentioning it so that someone could do further research on their own, is PUE or power use effectiveness, where industry leaders are nearing 1.0, which is using every watt of incoming power for computing, not waste. It just wouldn’t do for the author to mention that is possible to be very efficient. They get near a valid question to ask which is whether all the things we are doing on the Internet and in data center is worth the cost of the energy and other resources consumed, but they quickly move on to bashing rather than exploring that interesting question.
The most laughable part of the article is the constant harping on big, bad diesel generators, UPS batteries, etc (which industry leaders are moving away from anyway). The article comes close to implying that the datacenters run on generator all the time (uh, no), but later states “Even if there are no blackouts, backup generators still emit exhaust because they must be regularly tested.” So now we’re getting down to it, again, it’s clear the author wanted to write an article that “datacenters are big, bad, polluting, blah, blah, blah” so let’s not have facts or reason get in the way. Such things like that any datacenter operator wants their generators to run as little as possible, ideally only during monthly or quarterly testing. Nobody turns them on for the fun of it. The article again then goes after Amazon who apparently didn’t get some of the generators permitted in Virginia properly. Then for some reason spends a few paragraphs on minor permitting or other violations by some companies. It then says “permits had been issued to enough generators for data centers in his 14-county corner of Virginia to nearly match the output of a nuclear power plant.” making that sound like it’s a bad thing? Again anyone that knows the industry or does any real research knows that datacenters are typically built out in clusters (i.e. Quincy, WA, Northern Virginia, etc) so it should be no surprise that in those locations there are more datacenters. It also makes it sound like a bad thing that large generators have to be registered with local environmental agencies.
Throughout the entire article, the author makes it sound like the industry doesn’t care about these issues and that there has been no efficiency progress over the last 10 years. This is where the biggest disservice is done as it’s entirely possible to have very efficient datacenters, many already exist, and if there is real concern about lack of efficiency in existing datacenters, people and businesses have plenty of choices. Another simple search would have found Microsoft’s success here http://www.globalfoundationservices.com/posts/2012/september/12/microsofts-energy-and-efficiency-evolution-to-power-the-cloud.aspx or this one http://www.globalfoundationservices.com/posts/2012/august/9/microsofts-top-10-business-practices-for-environmentally-sustainable-data-centers.aspx
Why no interviews in the article with people that actually know what they are talking about? Most were with utility company people, or analysts outside the direct industry. It should have taken the author about 2 minutes to find James Hamilton’s (of Amazon) blog (http://perspectives.mvdirona.com/) and some actual knowledge about how much Amazon, along with Microsoft and Google are leading the industry in power efficiency. Or how any datacenter operator that knows anything knows that power is a major component of cost, and who is not trying to drive cost down?
The comments to the article are interesting, many of them are well informed and point out the obvious flaws. Others clearly indicate what the author and the Times was trying to do, get people all riled up and push for “regulation”. That’s the problem with bad and “gotcha” journalism, it energizes but does not inform. Shame on them for publishing something so silly and one-sided. They easily could have made the good point on the current levels of inefficiency in a lot of areas and the question of whether what we use technology for is worth the resource cost without taking all the cheap shots.