pg.ddx.io  pgsql-performance@postgresql.org mailing list archive  
help / color / mirror / Atom feed
random_page_cost = 1?
2+ messages / 2 participants
[nested] [flat]

* random_page_cost = 1?
@ 2005-06-09 16:37  Alex Stapleton <alexs@advfn.com>
  0 siblings, 1 reply; 2+ messages in thread

From: Alex Stapleton @ 2005-06-09 16:37 UTC (permalink / raw)
  To: pgsql-performance

Is this advisable? The disks are rather fast (15k iirc) but somehow I  
don't think they are covered in whatever magic fairy dust it would  
require for a sequential read to be as fast as a random one. However  
I could be wrong, are there any circumstances when this is actually  
going to help performance?



^ permalink  raw  reply  [nested|flat] 2+ messages in thread

* Re: random_page_cost = 1?
@ 2005-06-09 16:55  Tom Lane <tgl@sss.pgh.pa.us>
  parent: Alex Stapleton <alexs@advfn.com>
  0 siblings, 0 replies; 2+ messages in thread

From: Tom Lane @ 2005-06-09 16:55 UTC (permalink / raw)
  To: Alex Stapleton <alexs@advfn.com>; +Cc: pgsql-performance

Alex Stapleton <alexs@advfn.com> writes:
> Is this advisable?

Only if your database is small enough that you expect it to remain fully
cached in RAM.  In that case random_page_cost = 1 does in fact describe
the performance you expect Postgres to see.

People occasionally use values for random_page_cost that are much
smaller than physical reality would suggest, but I think this is mainly
a workaround for deficiencies elsewhere in the planner cost models.

			regards, tom lane



^ permalink  raw  reply  [nested|flat] 2+ messages in thread


end of thread, other threads:[~2005-06-09 16:55 UTC | newest]

Thread overview: 2+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2005-06-09 16:37 random_page_cost = 1? Alex Stapleton <alexs@advfn.com>
2005-06-09 16:55 ` Tom Lane <tgl@sss.pgh.pa.us>

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox