Re: Managing Large tablespaces

From: Howard J. Rogers <dba_at_hjrdba.com>
Date: Tue, 16 Jul 2002 06:27:24 +1000
Message-ID: <agvb63$m1t$1@lust.ihug.co.nz>

2Gb is actually 2048Mb, isn't it?

The difference is that setting 2000Mb is near enough 2Gb to be workable, but doesn't actually hit the proper 2048 number. 2048Mb is actually 2^32 (ie,

10000000000000000000000000000000) -so you've hit the limit of a binary

power. The next byte takes you on to the 2^33 range, and I presume that some O/S or file system code throws a bit of a wobbly crossing that 32/33 boundary. A bit like going from 99 to 00 caused some problems for some code.

Best to avoid boundary-sitting if you can, I think.

Regards
HJR "Steve S" <stevens_at_coloradocustomware.com> wrote in message news:bafba412.0207151156.180716f_at_posting.google.com...
> Pardon me for my ignorance, but I dont understand the diff between 2
> GB and 2000 M. I'm gueesing that it has to do with the the number of
> real bytes between the two numbers? So if so, how does the boundary
> issue create problems?
>
> "Paul Brewer" <paul_at_paul.brewers.org.uk> wrote in message
news:<3d308908_2_at_mk-nntp-1.news.uk.worldonline.com>...
> > "Steve S" <stevens_at_coloradocustomware.com> wrote in message
> > news:bafba412.0207130444.407a0c4f_at_posting.google.com...
> > > Richard,
> > > So you would recommend breaking up a large tablespace into many
> > > smaller (2 GB)datafiles for manageability. Is there a performance hit
> > > in spanning a tablespace across many datafiles over having one big
> > > file?
> > >
> > > THanks for your help,
> > > Steve
> > >
> > PMFJI.
> >
> > I agree with Richard, and moreover I tend towards the conservative. I
like
> > 2000M datafiles, rather than 2GB, as I'm slightly nervous of files which
end
> > exactly on a boundary.
> >
> > Just my 2c, and HTH,
> > Paul
Received on Mon Jul 15 2002 - 15:27:24 CDT