IOUG-A Live! 2002 - Day 5 (Thursday)
by Stephen Andert
Top 10 Tips for Oracle SQL Tuning by Dan Hotka had a lot of good information to use in daily tuning activities. He explained when nested loop joins may be preferred over merge scan joins. He showed a sample of tkprof output and what to look for. When considering adding an index, think about the impact it will have on insert and update performance. He had other good information and rading his whitepaper is a good way to get the information if you couldn't make it to this session.
Karen Kappler covered some of the new 9i Automatic Database Management Capabilities. Some new 9i features will be helpful to the new DBA's that get dropped into supporting a production system with less than complete training or in the case of relatively small or non-mission-critical databases i.e. Oracle Managed Files. Other features promise to take away some of the dirty work of a dba as using Automatic Undo Management instead of Rollback Segments promises to avoid the dreaded ORA-1555 Snapshot too old. I think real-world testing has yet to prove that but it still bears looking into.
Wei Huang had some interesting information in his presentation Tuning I/O for Better Performance. He explained why the SAME approach (Stripe And Mirror Everything) has weaknesses. He also explained why setting certain parameters like DB_FILE_MULTIBLOCK_READ_COUNT needs to be done with knowledge of the underlying file system. Again reading his white paper is a good idea for more information.
I'll wrap things up tomorrow with some notes and thoughts about the final sessions today and some overall conference notes.