Subject Re: Index is corrupt
Author svanderclock
> I would think it worthwhile to increase pagesize to 8KB. You have
> low selectivity (many duplicates) with deep index structures - this
> sounds (to me) to mean lower performance. How much "lower"? - I couldn't say.
>
> If testing proves that the low-selectivity indexes are beneficial - keep them.

yes we do lot of test and it's prouve it. for 90% of the query it's not help, but for 10% it's an unbelievable help ...



> As for (missing) Primary Key: I try to (almost) alway have a (32/64
> bit) Primary Key. There are exceptions, but I often regret those.

yes in theses 40 table i don't have any primary key (because i want to economize one more index)... you thing it's really matter ?