IBM unleashes Viper
Connecting state and local government leaders
Rapidly losing its sting in the database market, IBM Corp. hopes to turn things around with the release of the new version of its flagship DB2 database, code-named Viper.
Rapidly losing its sting in the database market, IBM Corp. hopes to turn things around with the release of the new version of its flagship DB2 database, code-named Viper during development. DB2 Enterprise 9, to be released later this month, has three major new features, according to product marketing manager Steve Miller.
The first two we'll go over quickly: One is labeled security, or fine-grained control of who is allowed to see what. This feature can be used, for instance, to keep database administrators from looking at the data in their charge, should they not have rights to do so. Another feature is improved compression, which can speed response time.
But the major feature of interest'one the company hopes will leapfrog DB2 over SQL Server, Oracle and all the open-source variants'is its ability to work natively with Extensible Markup Language-based files, something other major databases can't offer yet, Miller said.
'Native' is the key word here. Other databases work with XML files by decomposing or 'shredding' the document. Database engines must copy the data from the XML fields into corresponding relational fields. DB2 does not use this shredding technique, Miller explained. Instead, the system stores the XML files and searches them directly with XQuery, an SQL-like query language for XML documents. (Miller couldn't say whether the XML files were stored in a plain file system folder, within the database file itself, or within some other IBM container, so you might want to bring your storage administrator up to speed on these issues before a DB2 install.)
In any case, you can query across both relational databases and XML files. The software issues both a standard SQL query and an XQuery query, then aggregates the results. But won't XML searches be slower than those that use shredding? After all the whole point of setting up a relational database is that the data can be queried and ordered up much more quickly. Miller assured us that IBM has been working on speeding up flat-file searching for years'before XML even'and DB2 Viper draws on some of the resulting algorithms. It'll be interesting to see the independent benchmarks when they hit the streets.
For more IT trends and analysis, visit the GCN.com Tech Blog at www.gcn.com/blogs/tech.
NEXT STORY: Don't say 'J2EE'