Re: temp extent size

From: Maxim Demenko <mdemenko_at_gmail.com>
Date: Fri, 19 Jan 2007 21:23:34 +0100
Message-ID: <45B128C6.7050300@gmail.com>

joel garry schrieb:
> Maxim Demenko wrote:

>> joel garry schrieb:

>
> I think this may have come from an 8i dbca (though I don't think extent
> management local was there, really don't remember), or might have been
> 9iR2 :-O It reached it's maxsize long ago. I'm pretty sure I changed
> the sizes manually at some early point, and probably just ignored next
> since I knew LMT would deal with it however it wanted. If anybody
> cares, they can run dbca and see what it does. I was just pointing out
> to the OP that next makes no sense in O10, so will be ignored.

You were right on the point and it wasn't my intention to make impression i disagree with your posting.

> 1* select tablespace_name, increment_by from dba_temp_files
> SQL> /
>
> TABLESPACE_NAME INCREMENT_BY
> ------------------------------ ------------
> TEMP 80
>

Just looked on it on my test 10g database (default installation) - same picture. But i was probably wrong in assumption it would mean any performance difference (of course, it makes, but it should not account). Just traced the file resize for equally big datafile and tempfile - by datafile there is significant amount of 'Data file init write' and only 1 wait 'db file single write' in case of tempfile with following 2 'local write wait' (for 1st and 2nd blocks of tempfile). So, i can imagine the only a situation where frequent tempfile extension may be a problem - on the systems with high concurrency for controlfiles (as in both cases controlfile records will be written).

Best regards

Maxim Received on Fri Jan 19 2007 - 14:23:34 CST