Re: Performance of Views

> In a case like this the use of a view to define the join will probably
> help reduce the chance of bad SQL being developed to perform the same
> query since the view as coded should help ensure the best join
> order/method is in use.

Thanks, this is exactly my point. I've heard about compiling view etc. could you point me to some documentation.

Thanks.

Alain.

Mark.Powell_at_eds.com (Mark D Powell) wrote in message news:<2687bb95.0210011036.1f897543_at_posting.google.com>...
> akreienbuhl_at_yahoo.com (alainkr) wrote in message news:<9822b77f.0210010654.13456cb8_at_posting.google.com>...
> > Hello all,
> >
> > I need your advice for the following problem :
> >
> > I'd like to use a view for a query that is repeated througout our
> > code. That query has 7 joins. From a developer I'd really like that.
> > It would prevent each developer from aliasing with different name etc.
> > I'm affraid of performance problem since that view would be against a
> > quite large table 3,000,000 records / year
> >
> > A ran an execution plan for both (using the view and using tables)
> > there're obliously identical. Now does that mean that I won't have any
> > more performance problem with the view that without.
> >
> > I apreciate your help.
> >
> > Alain.
>
> Alain, no this does not guarentee that you will never have a
> performance problem related to the view, but it does mean you are
> probably OK.
>
> If you are using the CBO then outdated statistics or a permanent
> change over time in the relationship (relative size, data value range
> changes etc...) between some of the tables could affect the optimal
> plan and require revisiting the tuning process. This is also true
> with a straight query against the tables.
>
> In a case like this the use of a view to define the join will probably
> help reduce the chance of bad SQL being developed to perform the same
> query since the view as coded should help ensure the best join
> order/method is in use.
>
> IMHO -- Mark D Powell --
Received on Wed Oct 02 2002 - 03:57:17 CDT