Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: pctincrease = 0 or pctincrease = 1 is better?

Re: pctincrease = 0 or pctincrease = 1 is better?

From: Connor McDonald <connor_mcdonald_at_yahoo.com>
Date: 2000/06/26
Message-ID: <3957489F.3C90@yahoo.com>#1/1

Terry Dykstra wrote:
>
> What real life database has ever segments that you can size the same way?
> Tables, indexes come in every size, from very small to very big. Isn't this
> whole idea of identically sized segments wishful thinking?
>
> --
> Terry Dykstra
> Canadian Forest Oil Ltd.
>
> "Howard J. Rogers" <howardjr_at_www.com> wrote in message
> news:394c0ddc_at_news.iprimus.com.au...
> >
> > "Huy Vu" <huyv_at_usa.net> wrote in message
> > news:961124340.776256_at_news.aei.ca...
> > > Hi all,
> > >
> > > Our oracle database 8iv2 8.1.6 was created with pctincrease = 1. The
> > > result after running Oracle Expert program tell us the value of
 pctincrease
> > > should be changed from 1 to 0.
> > >
> > > I've never worked with pctincrease = 1 so I'm confusing now.
> > >
> > > Which value should I set for pctincrease will be better?
> > >
> > > Thanks in advance for any help.
> > >
> > > H.V
> > >
> > >
> >
> > Zero every time.
> >
> > A non-zero setting for PCTINCREASE causes SMON to automatically coalesce
> > free extents (ie, create large single extents of free space where before
> > there were multiple, small contiguous extents). Sounds like good news,
> > except that SMON will do its stuff when *it* feels like it -which may not
 be
> > quite what you want for performance reasons. Furthermore, nobody ever
 needs
> > coalescing if they ensure that all segments within a given tablespace use
> > the same size extents -since all your free extents would already be the
> > right size for all your segments, they wouldn't need coalescing into
 larger
> > ones.
> >
> > Given that a non-zero setting for PCTINCREASE will actually *cause*
> > inconsistent extent sizes, it's fairly obvious that PCTINCREASE actually
> > causes the problem it then goes on to solve, which sounds like a bad deal
 to
> > me!
> >
> > If you can't ensure consistent extent sizes for your segments, then you
 will
> > probably need some form of coalescing from time to time -but I strongly
> > suggest you do the 'good DBA' thing, and arrange for coalescing to take
> > place at a time of *your* choosing -by manually issuing the 'alter
> > tablespace X coalesce;' command (perhaps script all tablespaces to be
> > coalesced at a quiet time for the database, using cron or AT).
> >
> > In either event, it means that PCTINCREASE should always be set to zero.
> >
> > Regards
> > HJR
> >
> >
> > >
> >
> >

Nope...

All you need is several tablespaces (say)

BIG_STUFF

MEDIUM_STUFF

LITTLE_STUFF

etc

HTH

-- 
===========================================
Connor McDonald
http://www.oracledba.co.uk

We are born naked, wet and hungry...then things get worse
Received on Mon Jun 26 2000 - 00:00:00 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US