We added Rackspace calculations to our footprint data this week, as we’re keen to start running our Hadoop clusters over more infrastructure (in the greenest possible way) and we’re sure you are too.
I was getting a bit despondent about our sustainable options for running big computation, so was super excited to learn after talking to Rackspace and to Greenpeace that Rackspace have actually entered renewables purchase agreements for all the power in their London data centre, and that Greenpeace are also happy to sign off on those figures (Greenpeace are especially tough on greenwashing, so if they say it’s legit, we’re pretty happy). New ratings are here. The big green blob over the UK makes me smile.
Greenmonk just published a blog post about the same thing, praising Rackspace for being leaders in this area. We totally agree.
So Rackspace London rocket up the leaderboard as a green option, and we can build moar big data in the UK without doing bad things to the planet! Happy campers all round.
We show live footprint estimates for right now on our dashboard at http://www.mastodonc.com/dashboard, but it’s also pretty interesting to visualise the different locations over time to get a feel for the size of fluctuation to see how the horse race between time zones and temperatures plays out.
This chart shows some real ratings for May. It’s got some really interesting features.
Iceland (powered by geothermal energy) stays at a flat zero all the time.
Ireland has a pretty interesting profile – it’s cold enough to air-cool most of the time, although has a pretty crappy emissions factor on its electricity, so its footprint is not good but is stable (as long as it doesn’t get warm, which starts to happen at the end of the chart).
Sao Paulo is hot, so its footprint has a very clear diurnal cycle which tracks temperature and hence cooling power draw, but it’s on a good power source so the amplitude is small.
Yet another GigaOm article with a linkbait headline but genuinely interesting content, which is well worth a read – “The controversial world of clean power and data centers”.
It’s great to see this issue becoming increasingly mainstream, and in particular businesses starting to be more and more clear about the fact that being efficient is distinct from being green – and that clean power is a much bigger part of the carbon footprint battle than efficiency. They’re also completely on the money in pointing out that data centers need to run on baseload power (ie high and consistent levels) – which you can get easily from coal or from hydro or geo energy but not so easily from wind or solar, meaning that wind/solar powered DCs are not likely to be so realistic and scalable, even though solar panels on the roof are often put in for good PR.
The article does miss one major trick, though – it says that tech companies have to put their data centers near their customers, which is not always true. The exciting part about cloud computing and IaaS is that, for a lot of applications – like data processing – the latency to pretty much any well-connected facility in the world is low enough, which means that the data centers can follow the available green power and can be put down right next to baseload-type green power sources.
GigaOm published this interesting and provocative article last week, headlined ‘Why the days are numbered for Hadoop as we know it’.
It’s actually a pretty good argument, despite the sensational headline. The tl;dr, as I interpret it, is that: Continue reading
An interesting thought on waterfall, agile and long term vision came up on the London Java Community mailing list.
Richard Gomes asked, quite rightly, if agile and lean reject long term planning and vision. People who know me will know that I don’t agree.
For me, the problem with waterfall is that we end up waiting far too long to see if our original vision or strategy was worth pursuing.
Agile is all about figuring out the smallest, quickest thing you can do to test a vision. And if the vision is right, then that thing should start to return some value to the business. If the vision was wrong, then it tells us that we need to change the vision.