Re: Algorithm for calculating extent size in LMT

Jonathan Lewis doodled thusly:

>
>The key phrase, of course is "their own".
>It could make sense if you have been given a
>packaged solution and told not to muck about
>with tablespace names etc.

Ah yes!

>Some packages
>that we all know and loathe will then have
>1,000 tables in the tablespace, without storage
>clauses, and grow 5 of them to GB+ whilst
>leaving 800 of them untouched. In this case
>Autoallocate is a pretty good damage limitation
>strategy.
>

Absolutely! I can see it happening already, having a few of these "known" packages in my dbs. I shudder just to think of the claims by their "DBA experts":

"just define a constant extent size and use LMT, that's what Oracle recommends".

Remember my carping a while ago about the importance of "context" and making sure people don't go off on tangents with "silver bullet" recommendations?

I know it sounds repetitive and I've been on this now for nearly a decade. But given the slightest opportunity you can bet we're gonna have another wave of "expert recommendations" floating around that make no sense at all when taken out of context.

That's why I raised the flag. Not as an inherent problem with LMT.

Cheers
Nuno Souto
nsouto_at_optushome.com.au.nospam Received on Tue Mar 05 2002 - 03:37:15 CST