OLAP Goes Mainstream

OLAP expert Nigel Pendse spots leaders, laggards, and coming product breakthroughs.


Is OLAP a commodity now?

It's not fully commoditized—but it's well on the way. A lot of the niche vendors can't really compete with Microsoft-style prices. And even though their products may be good, they often require too much consulting and they're too hard to sell. They may be good to use, but the overall cost of ownership is too high. Microsoft doesn't have every feature, but the product is easy to buy, low cost, easy to install and works pretty well. It has fewer problems than the others, and for a lot of customers it does all they need.

Last year there was evidence that simple solutions were doing better than sophisticated solutions. So, simple reporting and dashboarding seem to be way more popular than the complex analytic functions like data mining, statistics and high-end OLAP.

Is it because there are more mid-market buyers?

It may just be that the sophisticated users are already well served. Even in big companies there are plenty of the less-sophisticated users that have nothing, and the priority at the moment is to give them something.

Analytic applications haven't grown as much as might be expected from the growth of OLAP applications....

Yeah, this is something a lot of people, particularly IDC's analysts, were pontificating was going to be a huge growth area. Although I didn't entirely agree with IDC, I also thought it was going to be a big market—and it simply hasn't materialized. The vendors that supply analytic applications charge a huge amount and the applications still end up needing a ton of consulting to deploy—so they don't really save people all that much. And they also impose on people a workflow and a style of management that a software vendor thinks is how people should run their companies, but maybe the people in those companies don't entirely agree. So analytic applications may be an expensive solution to a problem that the buying community doesn't think it has.

You've previewed OLAP products that many people are waiting for. What stands out?

The blurring of the distinction between OLAP and relational has been a long time coming, but there's real evidence that it's happening now. Hyperion's Project Avalanche came about to bring relational and multidimensional reporting into one. Cognos Series 8 does that. Microsoft SQL Server 2005 does it. Oracle to some degree does it by using SQL as the API for MOLAP cubes. It seems that, by a variety of routes, a number of vendors are finally delivering on an old forecast.

From the user point of view it's great because you don't have to care whether it's multidimensional or relational technology behind the scenes; you just want the information quickly. Generally, you need a mixture of the two, but as an end user you don't want to ever see the seams.

What should we expect from SQL Server 2005?

Handling larger dimensions, wide area network support, more complex dimensional modeling, many-to-many dimensions.... Those are just a few of the things that SQL Server 2005 does better than the 2000 version.


Recommended reading? In Search of Stupidity by Merrill R. Chapman. It's about many of the screw-ups in the computer industry over the past 20 years. Do any stand out? MicroPro threw away a winning position with WordStar, the first WYSIWYG word processor. You can't afford many mistakes when you're competing with Microsoft.

Editor's Choice
John Edwards, Technology Journalist & Author
Michael Gibbs, Founder, Go Cloud Architects
Jack Berkowitz, Chief Data Officer, ADP
John Edwards, Technology Journalist & Author