Teradata has taken the big plunge. Yesterday, as long rumored, it launched the first commercial solutions in its broad portfolio to be explicitly positioned as data warehousing (DW) appliances. Specifically, it announced the new Teradata 550P, Teradata 2500, and Teradata 5550 platforms, which join the established Teradata 5500 platform in the vendor’s market-leading enterprise data warehousing (EDW) solution family.
Previously, Teradata had been keeping its distance from marketing its solutions as appliances. Over the past several years, that buzzphrase has been popularized by pure-plays such as Netezza, Greenplum, and others to describe simple, low-cost, quick-deploy alternatives to the more expensive, complex solutions offered by Teradata and others. To some industry observers, the phrase bears a stigma, connoting feature- and capacity-limited hardware/software combos that are not ready for full EDW deployment.
One might argue that this is all semantic hair-splitting, and that the recent announcements are nothing fundamentally new for Teradata or for its customers. After all, Teradata effectively established the DW appliance market a quarter-century ago when it rolled out the first in a long line of preconfigured, preoptimized solutions that combine CPUs, storage, software, and database to address the most demanding analytical and decision support requirements.
But, in fact, yesterday’s Teradata announcement is significant on several levels: for the vendor itself, for the market as a whole, and for the DW/BI professionals who make up its core customer base.
For Teradata, the most strategically significant of the new appliance products is the 2500, which it is marketing as a entry-level EDW and analytical "sandbox" data-mart. The Teradata 2500 is priced and positioned for midmarket prospects who have turned to rival appliances rather than plunk down the megabucks needed for a high-end Teradata 5500 EDW platform. Even some of Teradata’s existing accounts have occasionally deployed, say, a Netezza here or a Greenplum there as a cost-effective OLAP accelerator for data marts in an environment anchored by the venerable Teradata 5500.
For the DW market as a whole, Teradata has effectively validated the appliance as the dominant go-to-market, packaging, pricing, and deployment model for full-fledged EDWs. In that respect, Teradata is simply following the leads set by EDW stalwarts IBM, Oracle, Microsoft, HP, and SAP in their earlier, more aggressive forays into this new territory.
Just as important, Teradata has essentially revamped its entire product portfolio as a family of DW appliances. With yesterday’s announcement, it has launched a tiered, appliance-based EDW product family that addresses a wide range of enterprise customer segments and price-points. Even its established 5500 platform may be regarded as a high-end appliance in Teradata’s new go-to-market strategy. The family ranges from a single-node SMP departmental DW or analytic sandbox (the 550P, at $67K per usable TB, maxing out at 6TB), to the entry-level EDW (the 2500, at $125K per usable TB, maxing out at 146TB over a 24-node MPP shared-nothing deployment), to the high-end 5500 and 5550 (each $200-250K per usable TB, with former scaling up to 1 petabyte, and the latter up to 4 petabytes, over 1,024-node MPP shared-nothing deployments).
No one should think that Teradata is merely dabbling in the DW appliance market. Though positioned for entry-level midmarket requirements, the new 2500 platform is nothing less than a full Teradata-powered EDW in a box. Indeed, all of the vendor’s new solutions incorporate the same mature, robust set of EDW technologies (i.e., Teradata 12 analytic database, query optimization, fast loading, workload management, parallel processing, real-time, data modeling, etc.) at the heart of the 5500 (though each of the new models differs slightly in such matters as range of operating systems, interconnects, processor cores, and high-availability features supported).
Of course, Teradata competitors are already scoffing at the 2500, pointing out that it’s far more expensive on a per-usable-TB basis (the DW appliance industry’s lowest common denominator) than feature- and capacity-comparable solutions from Netezza, Greenplum, DATAllegro, Dataupia, ParAccel, and others. It’s even more expensive than IBM’s InfoSphere Warehouse midmarket-focused C-class appliances.
Clearly, Teradata is still very much a high-end EDW vendor. In this regard, its closest direct rival in the DW appliance market is IBM, in terms of these vendors’ ability to provide full-featured "EDW appliances" for various price-points, with a clear migration/upgrade path from entry-level to high-end offerings. Of course, IBM can one-up Teradata by providing, in its InfoSphere Warehouse editions, access to data integration, quality, governance, and metadata solutions (through the IBM Information Server suite) and also BI tools (from its own Cognos or SAP/Business Objects). But Teradata can respond by pointing to the broad range of ETL, BI, data mining, and other partner solutions that integrate out of the box with its EDW appliances.
For DW/BI professionals, the significance of Teradata’s recent announcements is clearcut. Existing users of Teradata 5500 who are looking to offload some OLAP, reporting, and departmental data mart requirements to a more cost-effective platform should explore the 2500 (and the 550P for an analytical sandbox for development and data mining). Midmarket customers who have heretofore excluded Teradata from their DW appliance short lists should give it a fresh look.
Teradata’s entry into the appliance market provides information and knowledge management (I&KM) professionals with new DW choices, and is definitely not a "Netezza-killer." Users who have invested in competing solutions need not worry that their vendors will soon be pushing up the daisies just because Teradata has gotten serious on appliances. Teradata’s offerings have their pros and cons, and are not necessarily a "slam dunk" sale to its established EDW customers. And the vendor will undoubtedly face intensified challenges both from the pure-plays and from established EDW rivals.
As noted above, Teradata’s new low-end appliances are still considerably more expensive on a per-usable-TB basis than most of the competition. Also, Teradata’s DW appliances don’t include bundled ETL and BI software, which customers will still need to acquire from other vendors. Apples-to-apples (or, should we say, appliances-to-appliances) comparisons are still tricky in a market space as complex as this.
Nevertheless, Teradata’s appliances do have the advantage of enabling smooth migration of data, data models, table strucures, views, queries, load jobs, applications, and other artifacts to its higher-end EDW platforms as customers’ requirements push into petabyte territory. Likewise, customers’ DBAs, system administrators, and application developers will be able to leverage their existing skills across new Teradata appliance-based platforms.
That’s a scalability, life-cycle cost-control, and investment-protection advantage that the pure-play DW appliance vendors cannot yet match, here in mid-2008, though many of them are in the process of scaling their product architectures into the petabyte range.
Teradata’s entry into this arena signals that appliance-based EDW deployment is well on the way to becoming a best practice for information and knowledge management professionals. It is now possible to build a complete EDW from these high-performance building blocks.