Oracle Issues First Upgrade Of Sleepycat's Berkeley DB

Release 3.0 of the Java edition includes the ability to store Java objects quickly and cleanly to speed transactions and perform hot backups.
Oracle acquired its first open source product, the Berkeley DB embeddable database system, in February, and "the integration of the two companies is going well," says Mike Olson, former CEO of Sleepycat Software and now VP at Oracle.

The evidence: Release 3.0 of the Java edition of the Berkeley DB, which arrived on schedule last week during JavaOne in San Francisco. The release includes the ability to store Java objects quickly and cleanly to speed transactions and perform hot backups. All those features were in the plans for Release 3.0 before the acquisition, as reported in "Oracle Buys Sleepycat; Is JBoss Next?" in our Feb. 14 issue. Indeed, says Olson, there hasn't been enough time since the acquisition for the Sleepycat development team to alter Release 3.0 plans or assimilate some of the support resources available from Oracle, such as producing versions of Berkeley 3.0 that are "localized," or made available in different languages. In the future, Berkeley DB will be tapping into such resources. Oracle, he says, "has a lot of expertise" when it comes to executing product releases, including localization.

Oracle needed an embeddable system, one that can be combined with an application that can provide automatic database services, and Berkeley DB fleshes out the firm's product line, says Rex Wang, Oracle VP of embedded systems, which includes the TimesTen in-memory database system and the Oracle Light mobile system.

Release 3.0 can store simple Java objects, also known as pojos (plain old Java objects), with its new Persistence API, which provides a standard way for storing and retrieving them. In the past, Berkeley DB customers needed to write code linking the application to the object needed. The goal of an embedded database is to use links to retrieve needed data. It doesn't rely on SQL statements inputted by the user.

With the new Persistence API, the Berkeley DB doesn't rely on breaking objects down into rows and columns of relational data, followed by reconstruction at a later date. Instead, it uses the new API to store and retrieve them as simple Java objects, a process that's faster than an object tear-down-and-build-back-up procedure.

Wang says the 3.0 version can speed up transactions by adopting a deferred write mode, delaying writing the record of a transaction to a disk drive. The deferred write mode allows software to temporarily function as an in-memory database, speeding up critical transactions. In-memory systems run in a computer's random access memory without resorting to retrievals from or writes to disk. The Oracle TimesTen In Memory system, acquired last year, performs in-memory transactions on a much larger scale. See the Oct. 12 story, "Oracle Upgrades Its Recently Acquired TimesTen In Memory Database."

In addition, the 3.0 release includes the capability to invoke a hot backup helper for large databases to ensure constant backup records. The hot backup helper optimizes backups to ensure consistent, up-to-date information, Wang says.

Berkeley DB is available for free download as an open source system under the Sleepycat Public License, or as a commercially licensed system for distribution with commercial applications.

Olson says his fellow co-founders of Sleepycat, Margo Seltzer and Keith Bostic, remain active in its development, and the Sleepycat development team remains intact inside Oracle.

Editor's Choice
Samuel Greengard, Contributing Reporter
Cynthia Harvey, Freelance Journalist, InformationWeek
Carrie Pallardy, Contributing Reporter
John Edwards, Technology Journalist & Author
Astrid Gobardhan, Data Privacy Officer, VFS Global
Sara Peters, Editor-in-Chief, InformationWeek / Network Computing