Re: Clock sweep not caching enough B-Tree leaf pages?

Tom Lane <tgl@sss.pgh.pa.us>

From: Tom Lane <tgl@sss.pgh.pa.us>
To: Peter Geoghegan <pg@heroku.com>
Cc: Robert Haas <robertmhaas@gmail.com>, Andres Freund <andres@2ndquadrant.com>, Merlin Moncure <mmoncure@gmail.com>, Pg Hackers <pgsql-hackers@postgresql.org>
Date: 2014-04-16T23:29:00Z
Lists: pgsql-hackers
Peter Geoghegan <pg@heroku.com> writes:
> My immediate concern here is getting recognition of the importance of
> weighing frequency of access in *some* way.

That's a completely content-free statement; certainly the existing
clock-sweep code is sensitive to frequency of access, as would be
any other algorithm we'd be likely to adopt.  It may well be that
we can do better than what we've got, but sweeping generalities
are unlikely to help us much.

			regards, tom lane


Commits

  1. Replace the BufMgrLock with separate locks on the lookup hashtable and