Around a year ago, I wrote a blog on a publication by Hivos, an international organisation that seeks solutions for persistent worldwide problems, where datacenters were written off as polluters. Apparently, datacenters are an annual subject at Hivos: they published another article on their site on how bad datacenters are for the environment. This time even more biased than the previous article.
For whatever reason, Hivos has decided that datacenters are bad and is publishing articles annually to support that opinion. Whether the information is correct, does not seem to matter. Time for another reply.
Increase of 50 percent
Hivos's article starts by stating that data traffic is increasing with 50 percent annually on a worldwide basis and that 'these datacenters' are devouring energy. Data traffic is, obviously, not the same as datacenters. The equipment needed to transport all this data, is generally using very little energy, especially in comparison to the entire datacenter industry.
The servers in the datacenters are partly used for serving all kinds of content on the internet, but - and here I will use our own datacenter as a benchmark - to a very large extent on other things. Internal automation for example. Rows of servers that were previously not situated in a commercial datacenter, but in a datacenter owned by the company itself, or even just somewhere in a meter box.
There, they used just as much energy, often even more, because there is a lot less attention for the efficiency of cooling and emergency power on the smaller scale. Or servers that process email. What would be better for the environment, sending a letter on a dead tree, or an email? Quite the hoot is that the servers that are responsible for the services that produce the most data traffic, according to Hivos, are not even located in the Netherlands for the most part (Facebook, Twitter), and the part that is here, has not been included in the research (Google/YouTube).
Leaving a mark
When internet video's are concerned, only YouTube and pornography are mentioned. That indicates that Hivos is trying to paint the data traffic in a certain light. Of course, both are application that generate quite some traffic, but especially when discussing the differences with last year, it would be fair to mention Netflix for example. Or the fact that people (often without knowing) watch TV over the internet.
In short, commercial datacenters are not housing the equipment producing all the data traffic Hivos is so worried about, but many other servers. Servers that were previously located somewhere else and were not counted as a 'datacenter'. You cannot say that datacenters are providing and increasingly larger CO2 emission, because even though they are using more power, that is still largely power that was previously used somewhere else.
Next to that, datacenters are often a lot more energy efficient in cooling and emergency power installations and servers in datacenters are more efficiently used due to virtualisation.In short: the Dutch datacenter industry is doing well, but Hivos thinks it necessary, for the second year in a row, to write off the entire industry as a pollutant. Unjustified.