In-Depth

Analysis: A Tale of Two BI Strategies

IBM and Oracle approach BI from two different perspectives

• By Stephen Swoyer
• 06/18/2003

Third-party tools can exploit DB2 Cube Views to understand the underlying OLAP model of data that resides in Big Blue’s DB2 database.

With this new software, IBM appears willing to settle for the database portion of the BI stack, even as it cedes the market for BI front-end tools to specialty vendors. In this respect, Big Blue’s strategy is quite dissimilar to that of database market competitor Oracle Corp., which markets a line of BI products that are designed for the decision support tools space.

Anant Jhingran, director of business intelligence development with IBM, says that his company’s strategy has been largely determined by the realities of the marketplace. “[Customers] don’t expect to do rip-and-replace [of their existing BI front-end tools]. They have all of the technology investments that they need. Many of our larger customers have already invested in the Cognoses and the Brios and the Hyperions of the world,” he comments.

Enter DB2 Cube Views (formerly code-named “Aurora”): software that allows organizations to define OLAP-centric meta data—including hierarchies, dimensions, attributes, and business rules or formulas—and store it as a transparent extension to DB2. Cube Views OLAP meta data is accessible from a variety of different BI tools—from Cognos Inc., Brio Software Inc., and Hyperion Solutions Corp., among others—which can run directly against DB2 and process dimensional data without first importing it into their own proprietary cube structures, Jhingran says. “We do not want to get involved in battles between OLAP and data mining vendors. We want a full ecosystem of partners, and we want to provide a single version of truth for the data warehouse,” he concludes.

Exploiting 9i as a Lynchpin for BI

That may be so, allows Keith Gile, a senior industry analyst with Forrester Research subsidiary Giga Information Group, but IBM’s strategy has also been shaped by archrival Oracle, which in the past has leveraged the dominance of its Oracle 9i database to muscle into the decision support arena. The result, says Gile, is that Oracle enjoys a formidable out-of-the-box advantage in this space: “Oracle has such a stranglehold on a certain segment of the database world—particularly in relational databases and data marts, along with some operational applications—that they can play almost any game they want to play and still maintain leadership in a database capacity.”

Culturally, Oracle’s aggressive, compete-at-all-costs sales strategy mirrors the aggressive, compete-at-all-costs philosophy of its co-founder and CEO, Larry Ellison. In the BI space, especially, Oracle has not been afraid to aggressively compete against many vendors—such as Business Objects SA and Actuate Corp.—with which it has also partnered. “Oracle competes with its partners and wants their business,” comments Mike Schiff, a principal with data warehousing consultancy MAS Strategies. “Oracle thinks its install base is its own, and doesn’t appreciate its partners going in and taking share that it thinks it deserves.”

As a result, the database giant markets a full-blown BI suite, which includes a reporting component (Oracle 9i Reports); an ad-hoc query, reporting, analysis, and Web publishing tool (Oracle 9iAS Discoverer); an ETL tool (Oracle Warehouse Builder); and a portal (Oracle 9iAS Portal).

Oracle is also a seasoned OLAP player, having acquired a standalone OLAP server—Express OLAP—from Information Resources Inc. in 1995. When in 1998 Microsoft made the decision to drop an OLAP engine (dubbed “Plato”) into SQL Server 7.0, Oracle was forced to play catch-up. As a result, explains John Entenmann, VP of Oracle’s BI products, his company made the decision to integrate an OLAP component into its flagship database, even though it already marketed a separate OLAP server. “We took the analytic features that you would normally find in a multidimensional engine like [Hyperion’s] Essbase, and moved those capabilities to the database, and this is key because it means that you no longer need to move the data to a separate environment to do analysis, you can do it right in the database,” he explains.

In this regard, Oracle’s approach to integrated OLAP is further illustrative of the database giant’s compete-at-all-costs strategy. While Microsoft encouraged third-party vendors to develop front-end tools to exploit Plato, Oracle designed its integrated OLAP component as a replacement for the stand-alone Express OLAP server. As a result, Oracle 9i’s OLAP engine is designed to work first and foremost with its own tools. “Those [OLAP] capabilities in the database are also integrated with our tools, so Discoverer knows how to get at and use those analytics, warehouse builder knows how to generate an environment to populate the meta data for those analytics. It’s a very tight integration, very strong, stronger than what everybody else has,” comments Entenmann.

IBM has employed a different tact. First, the computing giant has partnered with a number of vendors—including Cognos, Hyperion and SAS Institute Inc., among others—to provide BI products and services to its customers.

With its DB2 Cube Views announcement, says MAS Strategies’ Schiff, Big Blue has taken this strategy one step further. “IBM is really trying to please all of the people, all of the time,” he comments. “It has traditionally partnered with Hyperion, which it will continue to do, but this is a move to open up DB2 to any interested third-party vendor. So IBM is opening [DB2] up to everybody, leveling the playing field, and still making its partner [Hyperion] happy.”

Oracle: Odd Vendor Out?

At least one analyst thinks that IBM’s BI strategy is aligned with a trend that is sweeping across the BI industry as a whole. “Many BI vendors are moving rapidly away from storing data—Hyperion and Cognos are both gradually downplaying their OLAP databases in favor of OLAP interoperability,” suggests Wayne Eckerson, director of research for TDWI. The odd vendor out is Oracle, observes Eckerson, which “is focusing on bundling everything with the database and selling the database.”

This strategy has served Oracle well in the past, argues Giga’s Gile, and looks to be a sound bet for the future—so long as Oracle continues to be the dominant RDBMS in most enterprise shops. “Oracle’s products are good enough for most applications. In some circumstances, and within an all-Oracle domain, they are all that most customers need,” he asserts.

In terms of database revenue and market share, IBM and Microsoft have increasingly made inroads at Oracle’s expense, however. Database market share reports from both International Data Corp. (IDC) and Gartner Inc. are contradictory—IDC shows IBM still trailing Oracle while Gartner has IBM out in front—but both firms say that IBM is closing the gap.

Microsoft, meanwhile, has eroded Oracle’s share at the low-end. The software giant recently unseated Oracle as tops on its own Windows operating system for the first time in years. Moreover, Microsoft has made noises about moving into the decision support and front-end tools arena, announcing a reporting component—Microsoft Enterprise Reporting—that competes head-to-head with Oracle’s own 9i Reports, as well as with leading BI players such as Crystal Enterprise.

MAS Strategies’ Schiff downplays the reports from both firms, noting that IBM’s numbers, in particular, are buoyed by its strength in legacy environments, where it still derives revenue from sales of IMS and VSAM. “In non-mainframe environments,” Schiff says, “Oracle is still the undisputed king.”