Blog: Barry Devlin

« April 2007 | Main | August 2008 »

Wednesday, 30 July 2008

Reviewing the reviews of the DatAllegro acquisition

I've been meaning to resurrect this blog for some time now, but, hey! life gets in the way. But, the recent coverage of Microsoft's acquisition of DatAllegro proved to be the trigger to get me going, though. It's all about feeds and speeds, bigger volumes, faster access and cheaper warehouses. Debates about how this will help Microsoft move up in the market and how it will impact the other vendors.

That's all very well and good, but, excuse me, have I missed something? Since when did data marts become data warehouses? I know that the appliance vendors tend to label themselves as data warehouse appliances, but I thought we all knew that was marketing. Of course, any appliance will be part of a data warehouse system in the broader sense. But when you look at the features and strengths that appliances have, you can see that they are really data marts. Data "hypermarts" perhaps, but marts nonetheless.

By definition, a data mart is a subset of the data in the enterprise data warehouse that has been optimized for use by a particular set of users. Such optimization includes selecting the data needed for some set of business purposes and structuring it to allow the fastest, most appropriate query access for users. It's all about how you get the data out! Sounds to me like exactly what the appliance vendors emphasize.

On the other hand, the data warehouse focus is on getting the data in. How to cleanse and reconcile the diverse data. How to ensure the cross-source timing is right. How to create a model that reflects the needs of the wider enterprise. And finally to make the consolidated view of the business available to the users - usually through data marts.

So, does the Microsoft acquisition disrupt the entire data warehouse market, sending the large players into a spin? I doubt it. Building a real data warehouse will continue to be as challenging as ever, requiring the same strong integration and project management skills as before as well as the deep database integration and manipulation technology that only the big relational databases possess as of now.

Posted by Barry Devlin at 7:36 PM | Permalink