Articles & Content

Archives

GALIO - DB2 Index Advisor: How We Implemented It and What We Get From Self-made Expert Tool (E07)

Topic: DB2 for z/OS

Subtopic: 2009



DATE: 2009-10-6 (14:15 - 15:15)
SPEAKERS: Viktor Kovacevic (Hermes-SOFTLAB)

Relational database tuning is a complex process which requires various levels of competence, from system and hardware engineering to knowledge of business logic. Optimizing application query workload with selection of proper set of binary indexes that minimize query response time and consecutively the resource usage is known as the index selection problem (ISP). As a stochastic, biologically inspired search method suitable for finding near-optimal solutions in complex search spaces, a genetic algorithm is suitable for solving this problem. In this presentation, Genetic ALgorithm for Index Optimization (GALIO), we are presenting one possible implementation of index advisor expert tool and how we can use itís functionalities in application tuning. Operational testing and usage of GALIO tool on real-world databases shows a significant improvement of optimization results in comparison with the results obtained without specialized index advisor.

EXP. LEVEL: Beginner,Intermediate,Advanced

OBJECTIVES:

How to automate application development and database optimization with self-made index advisor.

Evolutionary approach in index advisor implementation.

We report the optimization results achieved with the GALIO index advisor on an e-banking system database (DB2 UDB and DB2 on z/OS)

Overall explain plan statistics functionalities for SQL workload

Conclude with the ideas for future work on index advisor tool implementation.



Click Here to Download

NOTE: These are only open to members of IDUG. If you are not a member, please CLICK HERE for more information.