Blogs

DB2 10 for z/OS

More details emerge about new features that can slash CPU costs

Have you heard about DB2 10 for z/OS yet? The announcement of the beta program and presentations from the IBM System z Summit road shows highlight many of the new functions and features, and the upcoming International DB2 Users Group (IDUG) conference should provide further details. All of this demonstrates that IBM continues to listen to its customers: DB2 10 for z/OS addresses performance, scalability, availability, security, and data warehousing with features that can improve operations and overall performance, and—most important—reduce overall CPU consumption and total cost of ownership.

Many of these improvements, especially in application integration, SQL, and XML, can slash CPU costs for existing applications. Specifically, the DB2 10 optimizer does more analysis when choosing between two indexes, and it compares access-path costs more closely when determining the best access for SQL statements with OR and IN predicates. DB2 10 also improves the handling of SQL that overflows the relative identifier (RID) pool limits by using the DB2 work file resources and avoiding a costly tablespace scan. Additional CPU-reduction features include improved parallelism, LOB/XML streaming capabilities, and better handling of SQL Stage 2 predicates. These improvements will immediately benefit almost every application, without requiring any application changes or redevelopment.

DB2 10 also introduces a completely new access method called Hash Access, which uses a new hash space that provides direct access to a data row. In some cases, this direct access reduces data access to a single I/O, dramatically decreasing CPU use and speeding up application response time. There are trade-offs for using Hash Access, though: parallelism is not available and traditional clustering keys are not allowed against the hash data. Nevertheless, Hash Access will be great for product or customer identity data, and for other information where unique direct keys are already used.

As DB2 databases continue to grow in size and transaction volume, the amount of administration time required for database changes becomes more of a challenge. A number of DB2 10 utility improvements can further minimize downtime during normal operational and database-change activities. One of the best new features enhances operations by replacing the old DROP/RECREATE and REBUILD indexes method of adding new columns to a table to ALTER the new columns into the table and then performing an online reorganization for the changes to take place. The new method of enhancing tables eliminates downtime and improves availability for mission-critical very large databases (VLDBs).

With DB2 10, DBAs will also be able to create or rebuild a non-unique index without any application impact or locking downtime. This will especially help newly installed applications by making it possible to quickly define an index that can improve SQL access and accelerate problem resolution.This enhancement alone almost instantaneously improves performance for any installed application.

I’m looking forward to the beta program, road shows, and IDUG 2010 North America conference to bring out more details about all of the DB2 10 features. It’s great to see how IBM keeps enhancing DB2 so that its customers can continue to drive some of the biggest, most complex, and most robust database, data warehousing, and transaction systems in the world.