Re: Coalescing and PCTINCREASE
Date: 1996/12/07
Message-ID: <58ah01$dsh_at_news1.voicenet.com>#1/1
In article <589heo$6k2_at_newsgate.duke.edu>, ghubert_at_acpub.duke.edu says...
>
>I've been following the thread on Coalescing and PCTINCREASE.
>There seems to be an implication that it is often desirable for
>PCTINCREASE to be 0. Could someone confirm if this is so and
>briefly explain why?
>
>Thanks for your attention,
>
>gene hubert
>
Well, consider this simple example. You have a table which has an initial
and next extent size of 500K, and pctincrease 50. The third extent will
be 750K, the fourth 1125K, the fifth 1687K... you can see where this is
going. Assuming a "normally fragmented" (perhaps a non-sequitor, I grant
you) database, it doesn't take long to exceed the biggest available
extent. In many cases you want to control the number and size of the
extents without having them increase on you - for example, when striping
across multiple OS files to take advantage of parallel query option.
Dave Kreines
Director of Member Services, IOUG-A
DBA, Rhone-Poulenc, Inc.
Received on Sat Dec 07 1996 - 00:00:00 CET