Design question: one big field angainst child table

#21

"Matthieu" <vertonique (AT) yahoo (DOT) fr> wrote

Quote:

Hi every one and thanks a lot for your answers.

As Bill said, I've reproduced the behaviour of a kind of GROUP_CONTACT
function as explains

here:http://www.oracle-base.com/articles/...Techniques.php

I've created the child table for test purposes and populated it with 3
lot numbers for each "session" row. The table was so populated with
around 5 million of rows. I've choosen the make a PL-SQL function with
a cursor (no objects or create_type features)
For a limited range of data (from the begening of the year, otherwise
about 10000 rows ), the cost is about 100 ms.

However, if I add an order clause to my select, the difference is quite
huge between calling the sub functino or not: 870ms versus 3.2 seconds.

My guess is that your sort_area_size is the default 64K and thus the sort is

sorting to disk. Make it a little larger (a couple of meg) and the sort
should take only a little longer than the no sort.
Jim

#22

thanks you all again.
I've given up on looking for the sort issue as anyway it will be done
by the application (client side).

My tests underlined use of a child table isn't very expensive regarding
to the custom computation that would have to be done with a managable
long string field.

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Design question: one big field angainst child table

comp.databases comp.databases

Re: Design question: one big field angainst child table - 05-09-2006 , 08:17 PM

Re: Design question: one big field angainst child table - 05-11-2006 , 03:25 AM