In-Depth

HyperRoll Now Enhances Relational Data Warehouse Performance

The latest release of HyperRoll also includes algorithms to boost OLAP performance

• By Stephen Swoyer
• 05/19/2004

HyperRoll Inc. made its name as a purveyor of software that enhances the performance of OLAP engines from Hyperion Solutions Corp. and Oracle Corp.

New version 4.0 release of HyperRoll extends this value proposition into the relational space. The upshot, officials say, is that HyperRoll 4.0 can help officials aggregate information in relational data warehouse or data mart environments as well as business intelligence (BI) software from vendors such as Business Objects SA.

That’s not to say that HyperRoll 4.0 altogether neglects the bread-and-butter OLAP market in which it first established its reputation.

“We know that relational data warehousing solutions, just from a volume of data perspective, are significantly larger than most OLAP cubes, so we’ve continued to enhance the performance of our environment for OLAP environments,” says Rich Ghiossi, vice president of marketing for HyperRoll.

To that end, HyperRoll 4.0 includes new algorithms that help boost OLAP performance for existing users. “[Customers are] going to be able to see as much as a 2.0 order of magnitude improvement over our 3.0 product,” Ghiossi explains.

HyperRoll works by loading OLAP and relational data into a so-called aggregation engine, effectively emptying out the original OLAP cube. Officials say this helps improve performance by reducing load times and increasing query performance.

Ghiossi says that there’s a disconnect between conventional BI and operational BI solutions, such that many operational BI products are troubled by poor query performance or user concurrency. As a sort of kludge fix, he says, data warehousing professionals often turn to so-called “summary tables,” which—he acknowledges—are a fine solution when used sparingly. “To solve this performance problem, they’ll build three or four summary tables, and that’s a great solution. But then they’ll find that they need to build 10, and that turns into a bigger problem, then pretty soon, they’ve got to build 30, and the maintenance problems becomes unbearable,” he maintains.

Enter HyperRoll 4.0, says Ghiossi, the data aggregation capabilities of which are ideal for just this situation. “It’s a lot more effective is for us to build all of that aggregate information in HyperRoll,” he argues. “Where you’d think about building 30 or 40 or 50 summary table that are very specialized to certain reports, we can build all of those basically in one environment. It looks like one view within the database, so the maintenance headache becomes almost negligible.”

This has the effect of increasing query performance for both OLAP and relational tasks, he notes. “We were really making the OLAP environments run so much faster, and now we can get to the heart of the problem, these operational data stores and data warehouses and accelerate them.”

Hyperion, however, is aware of this need and is working on a solution of its own, Essbase Aggregate Storage Option (its internal, but highly publicized, code name is Ukraine). This option is currently in beta and, when available, is expected to be priced inexpensively, or even given freely to Essbase users as part of their maintenance contracts.

The new release has broader BI ramifications. For example, says Mike Schiff, a senior analyst with consultancy Current Analysis Inc., HyperRoll 4.0 can exploit the aggregate-awareness capabilities of Business Objects, along with the heterogeneous services functionality of the Oracle database. But the applicability of HyperRoll’s underlying technology isn’t restricted to a combined Business Objects or Oracle stack, Schiff says: “[O]ther BI and database vendors offer comparable features and by forming appropriate partnerships with BI vendors and systems integrators, HyperRoll should be able to expand to many additional environments as well,” he writes.

Previous versions of HyperRoll addressed problems in OLAP performance that are common to many vendors in the space, Schiff says, and while Hyperion is taking steps to redress some of these issues the HyperRoll technology is a godsend to users of Oracle’s Express OLAP engine. “[I]t certainly meets a need for Oracle Express users. Oracle is probably not going to invest a lot of effort to improve Express performance, other than perhaps encouraging Express users to migrate to the Oracle database and its extra-cost Oracle OLAP option. Consequently, this is definitely a market that HyperRoll can cultivate,” he says.