Data Warehousing: Is a Consolidation Craze Ahead?

Some industry watchers, primed for a bloodletting, go so far as to predict an imminent round of DW market consolidation.

With the acquisition of DATAllegro, it's unlikely that Microsoft Corp. will become a data warehouse (DW) appliance player. If anything, Redmond will take an approach that competitor Oracle Corp. took with its Optimized Warehouse (OOW) program, which it announced last September: partner with hardware OEMs to deliver optimized warehouses, tweaked or adjusted sized for configurations or workload requirements.

Microsoft hasn't confirmed that this is what it plans to do with its DATAllegro assets -- Redmond plans to outline its official roadmap at its Microsoft BI conference in a few months -- but officials haven't discounted the possibility, either.

"While we do have some hardware devices, such as Xbox and Zune, we are not a hardware vendor," says Fausto Ibarra, director of SQL Server product management with Microsoft. "The acquisition of DATAllegro allows Microsoft to expand our incredibly extensive ecosystem of hardware partners and solutions with a new software offering we will bring to market."

That alone won't put Microsoft's and Oracle's respective reference architectures at parity, of course. There are feature and functionality differences, for one thing, along with differences of scale. In fact, Microsoft/DATAllegro will have something that OOW currently lacks: a shared-nothing, massively-parallel processing (MPP) capability. That will put Microsoft in Teradata Corp. territory. How Oracle -- or the rest of the DW industry -- reacts to this upping of the DW ante could make for an interesting 6- to 12-month stretch.

Some industry watchers, primed for a bloodletting, even go so far as to predict an imminent round of DW market consolidation.

To support that view, experts say Oracle has run "hot and cold" on pushing MPP in its own product lines. Usually it opts to push either of two alternatives -- scalability across large SMP systems or its Real Application Clusters (RAC) -- as alternatives to MPP. But RAC isn't shared-nothing as MPP is, and it also seems to scale best between small numbers of SMP systems (theoretically, one can expand MPP scalability by adding additional nodes; there's no limit).

Large SMP, on the other hand, has a correspondingly large price tag: the 32-, 64-, or 72-way systems that really allow Oracle to strut its stuff in the high-end data warehousing segment carry mainframe-like price tags. Conversely, DW systems from DATAllegro, Teradata, and other players typically tap small SMP systems (of two to four processors) yoked together with other SMP nodes in an MPP configuration. The difference is important: two- or even four-way SMP systems are available today at relatively affordable prices, thanks to pushes from both Intel Corp. and Advanced Micro Devices (AMD) Inc. to go multi-core.

Will Oracle snap up an appliance player to burnish its own bona-fides in the high-end DW segment? If not Oracle, what about Microsoft's other competitors -- IBM, SAP, Sybase, or even Hewlett-Packard Co.?

Consider what happened last year in the BI tools market. Once Oracle bought Hyperion Solutions Corp., the onus was on its application and middleware competitors -- namely, SAP AG and IBM Corp. -- to make acquisitions of their own. The result: Business Objects SA and Cognos Inc. fell like dominos.

Thanks to Microsoft's acquisition of DATAllegro, some experts suggest, the still-gestating DW appliance space will probably contract, too.

"This deal sets the stage for what will surely be a period of rapid DW vendor consolidation," wrote James Kobielus, a senior analyst with consultancy Forrester Research, in a detailed post-acquisition blog entry last week. Expect the next 12 months to be bumpy, Kobielus predicts, as Oracle and other would-be DW leaders jockey for position.

"Over the coming year, Forrester expects incumbent enterprise DW … vendors -- or those who aspire to that status -- to acquire any of the growing number of DW-appliance pure-plays on the market. We expect many EDW incumbents to acquire DW appliance pure-plays, both to scale and accelerate their existing solution portfolios, and also to address the growing midmarket for cost-effective modular solutions," Kobielus continued.

"Given that several DW appliance pure-plays [he specifically cites Greenplum and Dataupia Inc., in addition to DATAllegro] boast low acquisition cost per usable terabyte, it stands to reason that incumbent EDW vendors acquire these vendors outright rather than attempt to hit those price-points through time- and resource-consuming modifications to their existing solution stacks." For this reason, Kobielus predicted that Oracle, SAP, and HP will probably make similar DW appliance-like acquisitions.

Not everyone agrees with this position. Jill Dyché, a principal with Baseline Consulting (and a Teradata veteran herself), who thinks that Microsoft's move -- while disruptive -- won't automatically trigger a round of DW industry consolidation.

"The acquisition really is a savvy move for Microsoft, since it officially brings the company and its SQL Server line into the world of MPP," she observes. "While Oracle has historically run hot and cold in marketing its MPP capabilities, the majority of Oracle customers aren't there. I'd bet that there's a conference room in Redwood Shores where the coffee is flowing and the marketing buzz is about whether to turn the MPP lights back on, and whether to hang the appliance shingle."

That being said, Dyché doesn't believe the onus is on Oracle or any of its other competitors to make an acquisition to counter Microsoft's move.

"I think it might be a bit hasty to predict this as the beginning of some sort of appliance buying spree on the part of the larger vendors," she says. "IBM and HP could already claim they have an answer to data warehouse appliance technologies. At this stage, I'd see any acquisitions being as much about putting competitors out of business as about bolstering incumbent solutions."