Re: Oracle Myths

On Monday, February 9, 2015 at 4:59:17 AM UTC-8, are..._at_gmail.com wrote:
> onsdag 15. mai 2002 10.25.36 UTC+2 skrev Niall Litchfield følgende:
> > PCTIncrease should be as small as possible but non-zero to minimize
> > tablespace fragmentation.1% is a good value (from my OCP course notes though
> > not necessarily given by the tutor!)
>
> When I was DBA on Oracle 7.3.4 I always set pctincrease on tables (not tablespaces) to 1% to trigger autocoalesce on extents. Perhaps this was a myth too? I can't check anymore, our 7.3.4 database was shut down late 2011.

IIRC, it was a myth that would kind of fragment the table, not quite being fixed by the coalesc, especially with "swiss-cheese" fragmentation. And of course giving SMON more to do might be a risk at times. But it was certainly a reason one would want to go to LMT. After some time, the table would still get a pretty big NEXT(as after all, this was advice given for DW). If you would do something like delete lots of rows then exp/imp, you'd get the initial way oversized. And with DMT, "oversized" meant huge FET$ and suchlike extent maps in the data dictionary, which could lead to either rebuilding the database or resorting to unsupported methods. Strangely coded enterprise apps with their own myths on top of this surely wouldn't help.

I'm sure Howard Rogers explained this better at some point, but he removed most of his stuff from the web a while back. There might be something found on usenet or asktom somewhere with sufficient searching.

jg

-- 
_at_home.com is bogus.
http://www.itworld.com/article/2881457/five-years-later-how-is-the-oracle-sun-marriage-working-out.html