I wrote a program that reads a table with a million plus rows and
writes to another table. I understand that use SQL is the fatest way
to do this. However, I need to keep track what records fail and
implement some logic. So, I have to use pl/sql to do this. I tried to
use bulk collect within cursor to improve performance (maybe set a
limit to 100 rows a time). However, I was told by an expert not to do
this because this database need to be up 24 by 7. Is that right? I did
a quick check in the manual and I could not find any restrictions like
that. I only saw the restrictions like 'run the program in the server
side not the client side', etc.
Did I miss something in the manual??? I thought it'll only reduce
burden for Oracle datebase engine. Thanks.
Received on Fri Sep 19 2003 - 00:36:31 CDT
