Re: Correcting next extent problems

>While Jonathan is correct, I will venture to say that this entire >discussion is misguided. Contrary to what the manuals lead you to >believe, having more than 8 extents will not have any significant >impact on the performance of your system. The reason 8 is used is that >8 extents are tracked in the tblspace shared memory structure. When >there are >8, the tblspace tblspace page for the table will be checked. >For frequently accessed tables, this will cause the TT page to sit in >cache, so there is no great loss. For system tables, it is even less >of a problem, because they are read into *local* cache when sqlturbo >is started up, so accessing that data is really not an issue. >So I wouldn't bother trying to change those extent sizes. One word of warning: We had a persistent data corruption problem on a moderately large (megabytes, not gigabytes) table that was ultimately attributed to too many extents. The engine was allowing us to insert duplicate rows. I don't have version or case numbers handy, but we went from dropping and recreating the index to unloading and recreating the table. After multiple attempts at recreating the table, we decided to calculate and use the proper extent sizes and the problem then when away. Your mileage may vary. -- David Stockel Internet: David.Stockel@snide.com