| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Simon Riggs <simon(at)2ndQuadrant(dot)com> |
| Cc: | Kevin Grittner <Kevin(dot)Grittner(at)wicourts(dot)gov>, Alvaro Herrera <alvherre(at)commandprompt(dot)com>, decibel <decibel(at)decibel(dot)org>, Itagaki Takahiro <itagaki(dot)takahiro(at)oss(dot)ntt(dot)co(dot)jp>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: FSM search modes |
| Date: | 2009-10-01 16:55:03 |
| Message-ID: | 21044.1254416103@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Simon Riggs <simon(at)2ndQuadrant(dot)com> writes:
> Yes, as Tom points out, this must be done with bias away from the very
> end of the table.
> I meant that we should start from the beginning of large spaces and that
> we shouldn't assume that all space worth filling is at start of
> relation.
Right. One of the goals that FSM is trying to meet is ensuring that
different backends aren't trying to insert into the same page
concurrently, so as to reduce page-level contention. So for example
it'd be a bad idea to adopt the really-dumb strategy of searching from
the start of the table for each request --- it'd funnel all the backends
to the same target page. What we want is a behavior that is randomized
but tends to prefer pages towards the start rather than hitting all free
pages equally. The btree precedent makes me suspect that quite a weak
preference would be sufficient. So for example we might try resetting
the search to the start of the relation with probability 0.01.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Magnus Hagander | 2009-10-01 16:56:00 | Re: Rejecting weak passwords |
| Previous Message | Euler Taveira de Oliveira | 2009-10-01 16:45:49 | Re: Limit allocated memory per session |