Re: Performance (was: The New Slashdot Setup (includes MySql server))

> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > All the sequential catalog scans that return one row are gone.  What has
> > not been done is adding indexes for scans returning more than one row.
> 
> I've occasionally wondered whether we can't find a way to use the
> catcaches for searches that can return multiple rows.  It'd be easy
> enough to add an API for catcache that could return multiple rows given
> a nonunique search key.  The problem is how to keep the catcache up to
> date with underlying reality for this kind of query.  Deletions of rows
> will be handled by the existing catcache invalidation mechanism, but
> how can we know when some other backend has added a row that will match
> a search condition?  Haven't seen an answer short of scanning the table
> every time, which makes the catcache no win at all.

Good point.  You can invalidate stuff, but how to find new stuff that
doesn't have a specific key?

--  Bruce Momjian                        |  http://www.op.net/~candle pgman@candle.pha.pa.us               |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026