"The arrival of Microsoft's OLAP server will greatly accelerate the growth of datamarts. And business users who previously couldn't even spell OLAP' will clamor for OLAP datamarts of their own", says Michael Schiff, principal at Current Analysis Inc., a solutions integrator in Sterling, VA.

Other pundits share Schiff's opinion. For instance, market researcher International Data Corp. (IDC) expects Microsoft's forthcoming Plato OLAP (online analytical processing) server to increase adoption of OLAP in the middle market. Currently the worldwide market for multidimensional analysis (also known as OLAP) tools is worth approximately $515 million, according to IDC. And that segment enjoyed a 36-percent CAGR (cumulative annual growth rate) between 1996 and 1997.

Relational databases do a great job of facilitating data-entry, efficiently storing data, and producing standardized reports. But relational databases choke when users request ad hoc reports in which three or more data types must be processed simultaneously.

OLAP databases solve this problem by replacing the column-and-row indexing schemes traditional used by relational databases' with multidimensional arrays in which each field of data gets its own dimension. This feature lets OLAP databases perform complex ad hoc queries in a fraction of the time that it takes relational databases to do so.

Consider, for example, a sales-analysis program that collects monthly data on products, regions, salespeople, and sales (five dimensions). Say that in a given month one salesperson outsold all of his or her counterparts. Crunching all the data to figure out why using traditional relational database tools is bound to be tedious process. OLAP makes this sort of complex analysis a snap. In this case, OLAP provides the tools to quickly drill down, find out which products each salesperson moved that month, and compare the results with the lists of products sold by the lesser producers. Because OLAP preindexes all variables, the OLAP tool would perform this sort of ad hoc query in the blink of an eye.

Roiling A Quiet Backwater

Teaming an OLAP server with a relational database provides the best of both worlds: stable, long-term data storage and fast, ad hoc retrieval. That's why all relational vendors field OLAP servers.

However, until recently OLAP was a quiet backwater in the software wars. But that situation changed in 1996, when the OLAP market really began to take off and when (probably not coincidentally) Microsoft elbowed into the OLAP space by purchasing well-regarded technology from Panorama Systems Ltd.

Since then Microsoft has worked hard to integrate its Plato product into SQL Server and BackOffice. By many accounts, the technology that Microsoft acquired was excellent, and the company has done a stellar job of product integration.

In June, Microsoft shipped 50,000 copies of SQL Server 7 (and the bundled OLAP server). If you're a solutions integrator who supports any kind of relational or OLAP database its now time to check out Microsoft's offering to see how (not if) it's going to have an impact on your business.

Another Paradigm Shift

If you want to be involved in the OLAP market, you'll need the following:

1. A deep understanding of the vertical market(s) in which you sell;

2. Software engineers who know how to make relational databases jump through hoops; and

3. An understanding of how to structure multidimensional databases and populate them with data from relational databases. This third requirement is a skill that most solutions integrators don't yet possess.

"Building a good OLAP structure isn't rocket science", says David Waugh, senior director of product marketing for Hummingbird Communications Ltd., an OLAP vendor based in North York, Ontario. "But don't assume that having good relational skills means you have good OLAP skills. They are different."

Microsoft knows that its solutions providers need new skills to help the company conquer the OLAP marketplace. Consequently Microsoft has pledged $20 million to subsidize the training required to sell and support the Plato/SQL Server 7.0 bundle.

Analysts predict three big survivors in this space: Microsoft, offering its Plato OLAP server; Oracle, with its Express OLAP server; and Hyperion Solutions and IBM through an alliance, contributing their Essbase and DB2 products, respectively. (Hyperion Software Corp. merged with Arbor Software Corp., developer of Essbase, this past summer to form Hyperion Solutions.)

Each of these OLAP products ties closely to a major relational database. Although the products can be cross-connected, most SIs would do well to concentrate on mastering all of the features of one major platform.

SIs with extensive backgrounds in data warehousing are likely to have all of the work they can handle selling and supporting pricey solutions based on Express, Essbase, or some other established OLAP tool. Thus, SIs that are new to the game will have ample opportunities to sell Microsoft-related OLAP solutions.

The Midmarket Magnet

"The arrival of SQL Server 7.0 provides an awesome opportunity to bring decision support to a vast new market that previously couldn't afford OLAP solutions", declares Richard Creeth, a principal of Creeth, Richman & Associates Inc., a datawarehouse consultancy in Norwalk, CT.

He notes that, to date, companies like Arbor Software have sold OLAP solutions primarily to Fortune 1000 companies with significant decision-support needs and budgets to match. The middle market is full of companies struggling with static reports, Creeth says.

Now that Microsoft is making OLAP affordable, you may want to consider OLAP enabling those database-driven applications you've already sold.

Adding additional reports to the vertical market solutions you sell and service is nothing new, of course. But Richard Daley, director of tools marketing at Hyperion Solutions in Sunnyvale, CA, says that OLAP tools make this job significantly easier.

Adding sophisticated sales analysis, budgeting, or forecasting capabilities using OLAP is an off-the-shelf sort of thing, Daley claims.

The price pressure that will likely ensue as a result of Microsoft's bundling Plato with SQL Server 7.0 should be used as a negotiating point to achieve pricing concessions from other OLAP vendors, advises Current Analysis's Schiff. You're in a particularly strong position if your existing OLAP vendor is concerned that you might choose Microsoft OLAP for some of your smaller apps, rather than use its product across the enterprise.

Joe Devlin is a partner in Armadillo Associates Inc., a solutions integrator headquarterd in Half Moon Bay, CA, in California's Silicon Valley. You may reach him at Joe @armadillosoft.com