DB2 re-emerges as a hybrid server

IBM’s 5-year development of the DB2 9 data server may offer technology that was worth the wait. In particular, DB2 9 can handle relational and native XML data, allowing DBAs to query across relational and non-relational data repositories at high performance rates using SQL, XQuery or XPath at the user interface. This hybrid capability—if it performs as promised in real world applications—has positive implications for data management, integration and service-oriented architectures.

“From creation to storage on disk and out into applications, we don’t monkey with the structure of the XML so that we can maintain the digital signatures and a lot of things that are unique to XML that you would lose if you forced it into the [relational] constraints that we used to do,” explains Jeff Jones, director of strategy, IBM information management software. While most orgs use XML databases, IBM is the first of the big three commercial relational database vendors to offer an internal storage mechanism and query support for what it calls pureXML.

Whether this technology performs outside of third party testing benchmarks and apps touted by IBM partners should be apparent in short order. IBM officially introduced DB2 9, aka Viper, for Linux, Windows and UNIX last week. The planned worldwide release date is July 28 with pricing at $4,874 per processor and $165 per user (minimum of 5 users) for DB2 9 Express. The zSeries version is slated for later this year.

“Data wrapped in XML messaging format is becoming a larger and larger part of the overall data in the corporation. By some measures it has reached 15 percent in large enterprises, so no matter how you slice it, that is a big deal for orgs to use it affectively,” says Wayne Kernochan, president, Infrastructure Associates.

“The IBM advantage in this case is really in the details. What it has done to improve performance is embed indexes and metadata specific to each type of XML data which should allow you to get that data off disc much faster. So far things like the Transaction Processing Council benchmark seem to indicate that this is true,” he continues.

IBM has traditionally had two ways of grouping data on disk: multidimensional clustering and hashing. A new feature in this release, range partitioning allows users to take a further step in achieving data locally, which should also boost performance, Kernochan says.

Although research suggests that people with existing applications using databases typically don’t switch, DB2 9 lends itself to new applications. For example, hospitals that need to store patient charts (relational data) and tests such as X-rays (non-structured data). Medical technicians would like to query the database and call up all of the information at once. This makes it easier, he says.

Other advances in this release include row compression for data objects, which promises to improve I/O performance, according to IBM. Beyond the existing encryption technology, there’s also a new way to apply row-level security by defining classifications and hierarchies. This feature is called Label-based Access Control. “Every row has a sensitivity applied to it and only people with the same sensitivity or greater can look at the data. It can be applied to the U.S. government standard or the standards in other governments,” Jones says.

DB2 9 also builds on IBM’s relationship, beefed up about a year ago, with German vendor SAP. Among the DB2 optimizations for SAP apps, is a silent install process. The companies have mapped out a roadmap for the next four releases of DB2 and SAP aligning development, release schedules and support agreements, says Jones.

“As databases continue to grow in capability and complexity at least internally, we must build in automation so that fewer and fewer resources are required,” he says. As businesses manage new types of unstructured information, application managers need to think of database systems more as data servers as less as traditional database management systems.

About the Author

Kathleen Richards ([email protected]) is the editor of and executive editor of Visual Studio Magazine.