Re: Algorithm for calculating extent size in LMT

"Ganesh Raja" <ganesh_at_gtfs-gulf.com> wrote in message news:a8aed4.0203060031.12152066_at_posting.google.com...
> > It's harder to predict in an autoallocate, since it
> > depends on what empty space actually exists
> > already, but there is a fair chance that your
> > starting segment would have
> > 3 x 1MB and 5 x 8MB.
>
> Sorry lost you here !!! What is the Logic Behind this.

PMFJI. Howard offered the following for the autoallocate algorithm.

"
The first 16 extents of a segment will be 64K in size. The next 64 extents will be 1M in size
Then extents become 8M in size."

I can't find the autoallocate algorithm documented anywhere (except that the docs for 8i agree that the smallest extent size is 64k which is a nice size). The logic though would be that you need at least 43mb and you have available extent sizes of 64k,1mb,8mb to choose from. 5*8mb + 3*1mb gives you 43mb in the smallest possible number of extents. This is different behaviour from that documented by Howard.

If I get a chance (or unless someone else jumps in first) I'll try to setup some tests later. If anyone can point the group at the documented behaviour or at some pre-done tests I think this would shed a great deal of light on the discussion.

It is also worth saying that as autoallocate has a variety of extent sizes to choose from over time the tablespace is going to become fragmented. I think this cost will outweigh the benefits of reduced data dictionary sql over time.

--
Niall Litchfield
Oracle DBA
Audit Commission UK
*****************************************
Please include version and platform
and SQL where applicable
It makes life easier and increases the
likelihood of a good answer

******************************************