Blog: Dan E. Linstedt

« Data Integration - Performance Numbers | Main | Does Big Data Equal More Business Intelligence? »

IBM - DB2 UDB 9.x - hot new technology

IBM is coming to the table with their RDBMS systems, finally making waves with their MPP options (labeled as DPF - data partitioning format), and multiple nodes. From a performance side of the house, scalability, and MPP are finally here. Not to mention new true XML interfaces, embedded XML within the RDBMS systems.

In their new 9.x DB2 UDB database, they are finally coupling MPP with hardware and software drivers that increase performance, scalability and manageability. They've got a good showing with their latest releases in BCU technology.

1. JUST ANNOUNCED: BCU for AIX Version 2.1 based DB2 Version 9 On December 6, IBM announced the BCU for AIX 2.1 - a refresh of the initial launch of the Balanced Configuration Unit for AIX in June 2005. This refreshed version of the BCU is based on DB2 Version 9, the latest p5 processors and the DS4800 Storage Subsystem. DB2 Version 9 offers some new features that will be applicable to many data warehouse implementations like table partitioning and data compression. More information about DB2 Version 9 can be found at: ftp://ftp.software.ibm.com/software/data/pubs/brochures/db2_9_overview.pdf

At one of our customer sites, the partitioning options are running at 300,000 rows per second on inserts - this is from a connected ETL tool. The performance is there, and no - the customer is not on a BCU, they are in fact, on Linux dual-core nodes. They are seeing linear performance increases when engaging multi-threaded connections.

DB2 V9 brings to the table the combination that makes big systems tick: hardware, coupled with optimized software algorithms to utilize that hardware, and additional partitioning options that allow optimizers to take advantage of data model designs - of course those designs must be built properly.

One of the hottest new features (I have yet to try) is the true XML (seamless embedding of XML information within the RDBMS) - allowing indexing, querying, updating, reading/writing / locking, etc... Seems to be very complete solution, and extends the Relational world into the semi-structured world. Well done IBM.

I would say that this release closes the gap between Teradata's technology and IBM's technology. *Note: Teradata was rated as the top right quadrant (leader in execution, leader in visionary) by a recent Gartner survey.

Now, if we could only get Updates and Deletes to be parallel / multi-threaded from an Application Programmers Interface (API) standpoint.... That would be awesome.

This technology doesn't come cheap - but its well worth the investment - especially if you are launching SOA, or incorporating unstructured data, or looking at DW2.0.

Cheers,
Dan L
CTO, Myers-Holum, Inc
http://www.MyersHolum.com

Share: