Re: hash join

From: Piotr Kolodziej <pkol_at_otago.gda.pl>
Date: Thu, 8 Apr 1999 10:07:05 +0200
Message-ID: <7ehodp$65r$1@korweta.task.gda.pl>

Andrew Babb wrote in message <370C5DB3.9967685B_at_mail.com>...
>It is quite possible for bad statistics to generate a bad execution plan.

That obviously is. After large and significant updates immediate re-analyzing may be necessary. But if tables are updated frequently, mainatinigs statistics would have significant overhead and effort. So it might be better to use rule based approach or lots of hints.

Unfortunetally, storing execution plans is available only in 8i.
On the other hand, it sometimes looks that the optimizer in 7.3.3 
& 7.3.4 tends to favourize HJ a bit if you have set choose mode.

--
Piotr Kolodziej pkol_at_otago.gda.pl
Just my private opinion. Received on Thu Apr 08 1999 - 03:07:05 CDT