Re: Tuning Question......

"Joel Garry" <joel-garry_at_home.com> wrote:
> Miggins wrote:
> > It cant really as the main cursor loop processes 20,000+ customers.

How does it know which 20,000 customers to process? Why can't it execute one query to get the data on all 20,000 customers, rather than execute 20, 000 queries?

> > For
> > each customer a sum of specific transaction details need to be
> > determined.

So? Isn't that what "group by" is for?

>
> That's why I suggested adding more specific indices. Both of the index
> range scans are probably picking up way too many irrelevant rows to
> scan through.

Unless something is very different than it seems (column type mismatches? drastic skew in the parent-child relationship?), I don't think that that is the case. One of the indices is using a fully specified primary key, so I really doubt that that one is picking up too many irrelevant rows. And since there is on average less than 2 children from every parent row, I doubt the other index scan is finding many irrelevant rows, either.

Xho

-- 
-------------------- http://NewsReader.Com/ --------------------
Usenet Newsgroup Service                        $9.95/Month 30GB