Virtualize This!Since E. F. Codd, databases have prospered with virtualization. Will it continue?Continued from Page 1 Linux the BarbarianOracle's approach emphasizes to its customers that taking advantage of the cost reductions presented by the combination of Linux and cheaper Intel-based hardware shouldn't include dismantling their data management investment in Oracle — in favor of open source databases such as mySQL. In other words, 10g is about splitting off the database management decisions from the commodity warfare going on underneath. Grid computing means different things to different people. But, to Oracle, it means a network of cheap hardware and storage that the DBMS can "virtualize" through RAC and related 10g features into a whole for easier data consumption and management. Giving users a "single Oracle database," applications using RAC "can dynamically leverage more blades provisioned to them," Goyal's white paper states. "Similarly, these applications can easily relinquish these blades when they no longer need them." Oracle asserts that this dynamism is not possible with shared nothing architecture, which DB2 on Unix, Linux, and Windows operating systems offers (although not exclusively). With shared nothing, "all the data needs to be repartitioned to allocate data to the new blades," the Oracle white paper states. "Similarly, when blades need to be taken off, data needs to be repartitioned before taking off the blades." IBM has both server and database management counterarguments to Oracle's approach. My main point, however, is that it is interesting to observe how the major vendors are trying to position virtualization in a way that is advantageous — and in a way that minimizes the impact of commodity, open source database systems. Based on audience comments during an October 8 panel session in which I participated (hosted by the San Francisco Bay Area Software Developers Forum, www.sdforum.org), interest in open source databases is growing. Many in the audience felt that open source commoditization was the inevitable future, and that the major database merchants were living on borrowed time. This opinion strikes me as a tad ahead of reality. Applications that require only basic DBMS functions are prime candidates for mySQL and other open source databases. No doubt, IBM would welcome such data resources into its federation — but it's doubtful that we will soon see an open source offering that could itself manage a complex federation, or even a grid full of blades. Open source systems will, however, put continuous pressure on the major vendors to rethink their value proposition and do more to help customers reduce cost in the entire database infrastructure. Demand PerformanceUtility, or "on demand" computing (to use IBM's term) will take virtualization to an even higher degree. Conceptual and logical architectures may know little about physical implementations, which could change as often as the owner of my mortgage does. The emphasis on business processes could further hide the significance of underlying database management. Will the "faster, better, cheaper" mantra that has ruled database management since the market exploded 20 years ago make vendors vulnerable to being virtualized out of relevance? Given trends toward more active, real-time systems, XML-based interchange, process integration, and new data types, database systems should have plenty of ways to differentiate themselves from the competition. That is, if they do not rest on their laurels. David Stodder [dstodder@cmp.com] is editorial director of Intelligent Enterprise.
|
 |
Most Popular This Week
IE Weekly Newsletter
Subscribe to the newsletter
| |
| |||||||||||||||||||||||||||||||
