Strolling around the exhibit floor at the TDWI conference in Chicago the past few days provided an interesting look into a rapidly evolving trend in data warehousing applicances. Of the 30 or so vendors exhibiting, I counted at least 7 that would be considered appliance vendors:
I might throw Oracle, HP, and Sand Technology in there as well, but I think you see my point - there seems to be the perception that there is a market for high performance "plug-in" systems to deploy data warehouses. What is perhaps even more interesting is that half of these vendor offerings are not specifically hardware appliances, but rather software database systems that can be deployed on top of different hardware systems - in other words, they are "software appliances" (!?)
In essence, many of these approaches, along with some from other vendors as well (Vertica was notably absent from this crowd, but showed up at the previous Las Vegas TDWI) focus on structural optimizations (such as columnar-oriented databases) that are very well-suited for loading into core memory and providing very fast read access, making it especially nice for query/reporting clients. The realization that the database system can be optimized and parallelized in a way that is decoupled from the hardware makes these software-only approaches look very cost-effective, especially when considering sizing a warehouse to meet current needs while considering future growth. Not only that, these systems are finely tuned for performance, (see Mark Madsen's comments about ParAccel's TPC-H benchmark scores).
The common theme with the software appliance crowd is lowering the barrier of entry to Small/Medium businesses seeking to jump on the BI bandwagon. WIth a variety of operational modes that span full-blown deployments (with hardwre purchase and integration) down to a service-based hosted model, this platform enables data warehousing at a fraction of the cost. This concept in its own right is worth some more exploration, and I think I may try to address that in an upcoming column.
Posted May 15, 2008 11:40 AM
Permalink | No Comments |