Re: LMT with autoallocate

There's nothing particularly wrong with 'autoallocate', except that one of the main reasons for moving to local management in the first place is that they can guarantee absolutely consistent extent sizes within the tablespace, and thus forever eliminate the possibility of tablespace fragmentation. Autoallocate guarantees precisely the opposite: a mixed set of extent sizes as the table grows, and hence a distinct chance of future fragmentation.

It's also the same argument about PCTINCREASE... it's a great saver for those who can't get their segments sized properly in the first place, but you *ought* to be able to size your segments properly!

Regards
HJR

--
Resources for Oracle: http://www.hjrdba.com
===============================


"Mark Bole" <makbo_at_ecis.com> wrote in message
news:3BEEA607.E4CC5761_at_ecis.com...


> I have seen several times the recommendation to only use uniform extents

with LMT


> instead of "autoallocate", but no explanation as to why?

>

> We have been using LMT with autoallocate for a long time (admittedly not a

huge DB)


> with no problems...

>

> Connor McDonald wrote:

>

> > Thomas Kyte wrote:

> > >

> > [...]

> > > >

> > > >Could someone who has used LMT (locally managed tablespaces) with

> > > >either autoallocate or uniform extents comment on their alleged

> > > >performance benefits, administrative issues, etc?

> > > >

> > [...]

> > >

> > > locally managed tablespaces are (with the glaring exception of system)

the ONLY


> > > type of tablespace I use.  I only use UNIFORM, never autoallocate.

> > >

>

> [...]

>