Re: HELP: select ... like speed issues

Glad to hear of success...

but this solution will run out of juice at some point, bringing up the 'need more hardware' question.

This might be way out there... but could you do an analysis of the existing structure of the text in the rows, and decompose it into smaller elements that could in fact be indexed and queried with the benefit of an index?

In other words, place the 'random' text into a list of more predictable textual components, like single words or small strings of words, then the main row ends up as a bunch of pointers to indexed text in another table which can be queried without the poreceding % on the like clause... definitely more up front work, and complicates the insert trigger.. but will scale well with any size hardware.

my 2 cents..
randy

"OnlineJobStore.com" wrote:

> Thanks, guys. Your feedback was really helpful. I've implemented full table
> caching w/running ANALYZE on startup so now the app works _really_ faster.
>
> Victor Urvantsev.
Received on Sat Jun 05 1999 - 18:05:25 CDT