In-Depth

An Appliance by Any Other Name…

A few weeks ago, Big Blue announced a new commodity data warehouse appliance based on 64-bit Opteron chips from AMD.

• By Stephen Swoyer
• 06/28/2006

Just over a year ago IBM Corp. announced its first-ever data warehouse appliance, a so-called Balanced Configuration Unit (BCU) based on its pSeries RISC servers and TotalStorage hardware.

Earlier this month, Big Blue refreshed and expanded its data warehouse appliance line. It was a not-insignificant update. After all, IBM’s first BCUs were RISC/Unix affairs—comparatively pricey pSeries servers powered by Big Blue’s Power5 microprocessors. (Although IBM did introduce Linux-based BCUs, these, too, were powered by pSeries servers.)

That changed a few weeks ago, however, when Big Blue announced a new commodity BCU—in this case, its Data Warehousing Balanced Configuration Unit 2.1 for Linux, an IBM eServer 326m system (which can be powered by up to two dual-core 64-bit Opteron chips from Advanced Micro Devices Inc.). Like IBM’s first-generation BCUs, the new BCU 2.1 appliance is dyed in the wool True Blue: it leverages IBM database (DB2 9 Data Warehouse Edition 9.1) and storage (TotalStorage DS4800) assets.

Of course, IBM still isn’t using the “a”-word—i.e., “appliance”—to describe its BCU bundles. But—as data warehousing consultant (and TDWI research collaborator) Mike Schiff puts it—if it looks like an appliance, acts like an appliance, and even conceivably smells like an appliance, isn’t IBM’s BCU offering, in fact, a data warehousing appliance?

Skeptics like Schiff have classified IBM’s BCUs as appliances by noting that they were based on proprietary hardware, used proprietary storage, and—at least with respect to the first BCU’s, which were an AIX-only play—ran on proprietary operating system software, to boot. That changed to some extent last August, when IBM announced new BCU packages based on Red Hat Linux and Novell Inc.’s SuSE Linux. The announcement earlier this month of IBM’s newest, almost-completely-commodity BCU bundles changes things even more. Now prospective customers can purchase a range of different BCU bundles, based on either proprietary or commodity hardware, and running proprietary or commodity operating environments. What’s so appliance-like about that?

Plenty, actually. Like explicit data warehouse appliances from Netezza and DATAllegro (as well as the high-end data warehousing systems marketed by NCR Corp. subsidiary Teradata), Big Blue’s BCU bundles are positioned, inescapably, as turnkey data warehousing systems.

IBM officials, for example, say the newest BCU bundles—like their predecessors—give organizations a modular way to build out their data warehousing infrastructures. The BCUs are pre-staged, pre-integrated, pre-tested and validated by an IBM services team. They’re also based on BI best practices IBM developed in conjunction with its large data warehousing customers. So, as Schiff puts it, once again: if it looks like an appliance, acts like an appliance, and smells like an appliance, isn’t the BCU—like similar offerings from Netezza (which uses a mix of pSeries RISC and commodity Intel servers) and DATAllegro (which uses an all-Intel approach)—in fact, a data warehousing appliance?

Semantics aside, Schiff says the appliance model has a lot to recommend it. But then, he concedes, that’s a no-brainer: Teradata has been a consistently reliable revenue generator for NCR, in spite of an atypically weak Q1 this year; Netezza rose from obscurity to data warehousing prominence in just four years; and upstart appliancer DATAllegro recently announced its first high-profile customer win (Sears Holdings), too.

The attraction, Schiff says, is turnkey expandability: As canned building blocks for creating and then augmenting a data warehouse, appliances promise to simplify many of the implementation and integration issues that traditionally bedevil data warehouse adopters. For IBM-centric environments, he concludes, Big Blue’s BCUs provide an alternative to the more platform-neutral appliances shopped by Netezza and DATAllegro, as well as the full-fledged data warehousing platform marketed by Teradata.