The next wave in BI
Beyond Big Data: the silent trends in IT & BI31/12/2012 13:50
As the end of 2012 is nearing, many analyts and market watchers are posting their insights on how 2013 will look like, what major trends can be seen and how we should prepare ourselves for those developments. And as always, if you have followed the general trade press, visited some conferences and attended a few webinars, you won't hear of lot of news. To some extent that's reassuring (the 'ha, I knew it' effect), or even boring ('nothing new happening'). In any case, it's been hard to miss that everything is (and will be for some time to come) about Big Data, Mobile, Cloud and Analytics (to name a few). These developments are hard to miss because they have surfaced some time ago and are now mostly on the top of their hype cycle. But what goes on beyond the surface? Which technology trends are less visible but are silently revolutionizing the way we work and live? I think there are at least three of these trends that form the undercurrent of recent developments and are already having a major impact.
1. Voice interfaces & natural language processing
We probably all remember Watson's victory in the US TV game show Jeopardy where questions were asked to a computer in plain English. Although that was a purpose-built multi-million dollar supercomputer which most people don't have lying around in their office, some of this amazing speech recognition technology is used by many people on a daily basis when they ask something of Siri on their iPhone, or ask directions on their Android phone. Currently this technology is mostly used to ask for a phone number, find a restaurant or get directions, but what would happen if you could combine Siri with a tool like https://www.neutrinobi.com/ which already allows you to ask questions about your data in plain English? In my view, the building blocks are all there, it's just a matter of time before someone will integrate them.
2. Connectivity & bandwidth
When I started my first consulting job in 1993 I didn't get a mobile phone or laptop. These things simply didn't exist back then, or were too expensive to hand out to mere mortals like myself. Nowadays we've almost reached the point where everybody is always connected. In 1993 I had a 14K4 modem which by the standards of those days was pretty fast. Being online was expensive since every minute meant an extra tick on your telco's invoice. Fast forward to 2013: fibre optic networks in the making, and 60Mb/sec flatrate internet is pretty common, at least in the Netherlands. The same will happen with mobile internet: flat rate based on speed and no (well, almost no) data limits. Mobile speeds will move towards the 50-100Mbit/sec range in the near term. What this means for BI? Always access to your data, always connected, no time and place constraints. It means liberation from our physical boundaries and less traveling (everyone can be reached instantly on Skype or Google Voice). It also means that we need to rethink how we work and live: does always online also imply 'always at work'?
System on chips (SOCs) have been around for some time since most of our mobile devices are powered by one. However, it's not those SOC's that interest me, it's the new breed of Servers on a Chip. A SOC is a complete computer molded on a single chip, utilizing a lot less power than tradional CPU's and GPU's. Where most PC and server CPU's run on an x86 based Intel or AMD cpu, most new SOCs are all based on an ARM (RISC) architecture. Low power usage doesn't necessarily mean low computing power though: since the ARM architecture is so efficient, the chips don't get as hot as regular CPU's, and many of them can run in parallel on a very small footprint. Combine that parallel processing at low power and low cost with direct access to local storage and you'll get the picture: an extremely efficient platform for handling large volumes of distributed data, whether this is Hadoop or any of the other (No)SQL database flavors. In 2013 we'll see the start of a major shift in the datacenter, where traditional x86 servers will be replaced or augmented by their much more efficient ARM 64 bit Server On a Chip counterparts from companies like Calxeda ($100 Mln in VC!) and Applied Micro. Those might look like obscure names now, but companies like HP, Dell, Facebook and AMD (to name a few) have already started positioning themselves for the ARM data center takeover.