Re: datafiles space allocation algorithm

  • From: "Niall Litchfield" <niall.litchfield@xxxxxxxxx>
  • To: amonte <ax.mount@xxxxxxxxx>
  • Date: Fri, 6 Oct 2006 12:26:27 +0100

What we do in this situation is purely to create datafiles with a standard
size (I usually pick 2049M)  and not set to autoextend, that way you'll
always be able to move them around if needed.

cheers

Niall


On 10/6/06, amonte <ax.mount@xxxxxxxxx> wrote:

Hi

The problem is my tablespace will have over 30GB and having 8 datafiles
where one has 2gb and other 1 and other 1 and one 26gb.... doesnt seem
reasonable



On 10/6/06, Christian Antognini <Christian.Antognini@xxxxxxxxxxxx> wrote:
>
> Niall
>
> > This is true for the straightforward case of one table on its own in
> > its own ts but is affected by factors such as
> >
> > other objects also growing at the same time
> > files having autoextend or not - especially if there is a mix in the
> > same ts
> > available 'holes' in the existing datafiles
> >
> > and so on.
>
> I agree with you. I was just trying to explain from where the
> "round-robin" stuff comes from... No more no less.
>
> > In general though I can't really see why, other than curiosity,
> > anyone would care about where a next extent would be allocated these
> > days. (just that it would be).
>
> One more, I completely agree. Such stuffs, i.e. spreading extents over
> many files, were performed loooong time ago to spread data over many
> disks. Today, usually, it makes no more sense.
>
>
> Cheers,
> Chris
>




--
Niall Litchfield
Oracle DBA
http://www.orawin.info

Other related posts: