On 2013-05-03 12:15:27 -0400, Alvaro Herrera wrote:
> Bruce Momjian escribió:
> > On Fri, May 3, 2013 at 05:24:54PM +0200, Andres Freund wrote:
>
> > > The problem with an extra metadata fork is that it essentially would
> > > double the files in a cluster and it would also noticeably increase the
> > > amount of open files we need.
> > > There have been quite some complaints about CREATE DATABASE speed, I
> > > am not sure we want to make it even slower :(
> >
> > Agreed. We start to get into file system performance issues at that
> > point. I have often wondered if we need to create hash the files into
> > subdirectories for databases with many tables. Has anyone profiled
> > this?
>
> Modern filesystems use trees to store file nowadays, not linear arrays,
> and so that old scenario (long time to find one specific directory entry
> within a directory containing lots of files) is no longer that serious a
> problem.
Absolutely. Also, we normally shouldn't open/close files constantly
since we cache open files so open(2) performance shouldn't be *that*
critical.
Greetings,
Andres Freund
-- Andres Freund http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training &
Services