Building a data warehouse is pretty easy, really. You capture a whole big pot of detail data (register scan data, inventory item data, individual banking transaction data, etc.) and keep it in a big online storage heap. It's all there.

Now, all you have to do is make sense out of it. And that's the tough part.

Most businesses don't have a clue as to what all that data means. Really. They know that they sell widgets like there's no tomorrow, but they don't know WHY people are buying the crap out of their widgets all of a sudden. And the reason is that they have very little idea of how their products or services are actually perceived at the sale point, or how the sales relate back to the way the rest of the business operates.

With a data warehouse, you have all the little detail pieces - like having hundreds of hours of video from surveillance cameras all over the factory. Unfortunately you can't watch everything at once, and it just looks like random motion.

Now, add a model for the business (and most businesses don't have a model that reflects their actual operations) that is realistic, and suddenly all those little random bits of data start making sense. Add in some external event data, like holiday schedules, demographic data, customer detail data, supplier data, transportation data, and the driving events that cause customers to rush in and buy plywood just before a hurricane hits start coming together.

And...the key tools for doing this are complex analytical programs that can troll all those random bits and start determining patterns. Once the patterns and relationships start emerging, you can go back to your business model and make the changes that bring it into line with reality. And, of course, once you get the reality check thing working, you can now start "tweaking" your business processes and see meaningful results from the tweaks. Not necessarily always positive changes, but now you're not guessing as to what caused the increase in sales of beer and diapers and why putting them together on the shelf did nothing to improve the sales further.

So, this is why Netezza is buying NuTech Solutions. Selling cheap boxes to replace stuff like Teradata and DB2 is only part of the picture - and a small, low-margin part at that. Being able to deliver the on-going BPR and analysis consulting is where the REAL bucks are, and, unfortunately for Netezza, the tools that Teradata and IBM have worked generations developing to support their customers data warehouse use are NOT transferable to the Netezza solution, and using off-the-shelf tools like Microstrategy or Hyperion provide the pretty pictures, but do little to actually automate the BPR analysis. That's still the task of high-dollar analysts and their proprietary tools and captive stables of mathematicians.

All in all, a good move for Netezza. Maybe now they'll get some real traction against the incumbent DW vendors.

Oh, and by the way: you're wondering what HP is doing buying EDS? This is the same situation as Netezza, only multiplied by all the other parts of business operations: data center optimization, workflow optimization, NeoView data warehouse (almost enough by itself to justify the EDS buy), ERP integration...the list goes on.

In both cases it's a cheap way to jump-start what would normally be 10 years of internal development and experience. Think about it.

Having done a masters in Machine learning a couple of years ago, there's nothing radically new here. Genetic programming dates back to the 1970's and Fuzzy mathematics goes back to the sixties.

So it's a company of people who've studied and apply AI techniques. The only reason this stuff is 'out there' is that the techniques can need fairly massive computation (I used around 30 processors for 3 weeks solid to generate the results for my thesis), and hence have not been practical until relatively recently.

I trust in Global Operating Devices, they are not pushing/pimping phished, as in stolen/purloined/filched, Intellectual Property from a Cloudy Server Source for IT which is Out of this World, for anything HyperRadioProActive is Deadly Toxic containing Astute NEUKlearer Triggers which Detonate Self Destruct Sequence Shifting with Information Overloads and DDOS, without the Advanced Permissible Link PoWeRControl Key.

"Also, their logo is a Sierpiński triangle. Maybe that should have warned us we were jumping into the rabbit hole here.

Austin, maybe they been chancing their arm jumping into rabbit holes which are Really Black Holes of Information and Intelligence, CyberIntelAIgently Designed to Test for Fraudsters and Free Loaders/Trojan Downloaders/Pirating Noders.

And Enigmas are always of Interest as Companies and/or Businesses/Virtual IDEntities for they are never what they seem, with some being more and others being less than expected.