Living in the cloud

Over lunch the other day with VirtualSharp (of which more at a later date), I was asked what I saw as being the hot areas (as it were) for the datacentre. I responded, predictably enough perhaps, that I saw the intertwined issues of power, cooling and efficiency as the most important issues that datacentre designers and managers will be focusing on for the foreseeable future.

But I should have said that it's all about storage -- bear with me.

I don't think those things -- power, cooling and efficiency -- are going away, although I am heartened by the micro-trend of modularity, which is starting to see datacentres being sold in chunks rather than monolithic facilities, and of more innovative approaches to cooling, including fresh air. When money drives, a solution can often be found....

And maybe we'll see an increase in the number of datacentres in cold places (like most of the UK) as a result, and fewer facilities being plonked down in what seem to me to be utterly bonkers locations such as Phoenix, Arizona, where average high temperatures are over 37 degrees C for three months out of the year, and have spiked at over 48.9 degrees C. Who would park a fridge there?

So what we didn't talk about much over that lunch was storage, in the sense of the volumes of the stuff we'll need, and why we'll need it. This is hardly a revelation but it got me thinking about the amount of storage my life will occupy somewhere in the cloud.

Rupert Goodwin's blog from Intel's developer forum indicates where things are going. As he points out, mobile phones are starting to store information about who we are, what we do and what we like. That data heads off to the cloud and sits there, awaiting analysis. Right now, we can do simple things, such as "using location, calendar, audio sampling and software usage to spot when you're in a meeting and, further, when you're actually giving a presentation and should not be interrupted by anyone, not even by IM or text."

In other words, it's a form of intelligence: reaching a conclusion based on evidential data. Right now, there's not a lot of data on us, but it is growing fast. With sensors galore in our phones - we'll see much more of this kind of feature in years to come -- our personal devices will know a lot about us.

But how will they make sense of it? One of the failings of artificial intelligence was not just a lack of computing power, but a lack of real-world data or common sense: computers can play chess, a problem with strict limitations, but still cannot recognise faces, a completely different scale of problem.

But at some point, those computers (smartphones) will know enough about you and your habits and be powerful enough to make sensible suggestions or even decisions about how to act or react, to the point where you or I will find them helpful. We're not there yet, but it seems close enough not to be science fiction any more.

What's all this got to do with datacentres? Only that the sheer volume of data needed to start making these decisions is truly vast. And when every data point of your life is up there in the cloud, on a spinning disk (or some NAND gate), then your devices will be able to start making sense. The implications for your storage footprint are vast, but there seems to no way of stopping it -- unless we finally run out of energy....

What brought it into focus was the launch yesterday of a new phone that offers a glimpse into that future: HTC's latest Android smartphone, the Desire HD. Its new HTCSense.com feature consists of a set of cloud-based services, mainly focused around issues such as email, text messages, remote control of the phone. Some of those services are semi-autonomous -- and this is just the start.