| From: | johnnnnnn <john(at)phaedrusdeinus(dot)org> |
|---|---|
| To: | pgsql-performance(at)postgresql(dot)org |
| Subject: | automated index suggestor -- request for comment |
| Date: | 2002-12-13 03:22:38 |
| Message-ID: | 20021213032238.GC8912@performics.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-performance |
The manual is pretty sparse on advice regarding indices. Plenty of
good feature documentation, but not much about when and where an index
is appropriate (except a suggestion that multi-column indices should
be avoided).
Of course, the ultimate arbiter of which indices are used is the
planner/optimizer. If i could somehow convince the optimizer to
consider indices that don't yet exist, it could tell me which would
give the greatest benefit should i add them.
So, i'm writing for two reasons. First, i want to gauge interest in
this tool. Is this something that people would find useful?
Second, i am looking to solicit some advice. Is this project even
feasible? If so, where would be the best place to start? My assumption
has been that i would need to hack into the current code for
determining index paths, and spoof it somehow, but is that possible
without actually creating the indices?
Any and all feedback welcome.
-johnnnnnnnnnn
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Hannu Krosing | 2002-12-13 12:05:36 | Re: automated index suggestor -- request for comment |
| Previous Message | Stephan Szabo | 2002-12-13 02:11:50 | Re: [PERFORM] CLUSTER command |