Re: fragmentation issues

"yls177" <yls177_at_hotmail.com> wrote in message news:c06e4d68.0405290510.40d41c4a_at_posting.google.com...

> > For DMT, you can *try* to help things along by specifying MINIMUM EXTENT
for
> > a tablespace. That is not part of the default storage clause, and
therefore
> > cannot be overridden by anything in the segment's specific storage
clause.
> > However, it still doesn't do the job, because it is *interpreted* in the
> > light of a segment-specific storage clause. If minimum extent is 500K,
and
> > your segment requests an initial of 23K, it will be given 500K. But if
the
> > segment then asks for a next of 612K, it will be given a 1MB extent...
at
> > which point, initial does not equal next for that tablespace. It
certainly
> > means you have less variation in extent sizes, however, and is therefore
the
> > next best thing to using LMTs.
>
>
> after the segment requests for an initial of 23k, 377k still remains
> in the extent. if another new segment requests for 70K, will it use
> the same segment as the 23K or a total new extent will be given?

Remember: an extent belongs exclusively to a segment. If one segment is allocated 500K, uses 23K and has 377K of free space in the extent still to go (though I think mathematics suggests it might be more like 477K), that has absolutely nothing to do with the second new segment. If it requests 70K, it will be given a brand new extent of 500K. Segments don't share extents.

> also, a 1MB extent is given because the new segment is asking for a
> 612K, whereas the minimum extent size is 500K, therefore it is not
> sufficient and hence 2 is needed or 1 extent with a size of 1MB? .
> which means to say that the extent size are not uniform which is
> fragmentation
> >
> > You can retrofit a DMT tablespace with a MINIIMUM EXTENT clause (alter
> > tablespace X minimum extent Y). But that doesn't do anything for the
extents
> > within the tablespace that have already been allocated, of course.
> >
>
> here, u were saying that the above sql command is not valid for those
> extents that have already been allocated?

It's not "not valid". It's a perfectly legitimate bit of SQL, that will be accepted at any time. But it won't cause the tablespace to re-size those extents which have already been allocated. They will still have their original sizes, even if they are extremely weird and have not the slightest resemblance to a new MINIMUM EXTENT clause.

Regards
HJR
> > Regards
> > HJR
Received on Sat May 29 2004 - 08:27:56 CDT