In the course of customizing one of our 3rd party apps that uses
Entity Beans to do multiple inserts to some tables, It has been
decided we'd use triggers on the tables to do some RI logic and
insert essentially duplicate data into another database - Oracle
Financials interface, to be exact,
 
I've been opposed to this methodology, on the grounds that
nobody here can tell me exactly how this app is behaving
in the first place. I've heard that some beans do stupid (sic) things like
insert a row, then update each column in the course of a single
transactional insert, thus requiring both insert and update triggers;
performance could be nightmarish, assuming we don't hit 4091
errors even trying to write the code.
 
Unfortunately, without some hard facts, I'm going to have a 
hard time convincing anyone this is bad practice. And
without tools to see exactly how the object(s) behaving, I won't
be able to prove it on our development platforms either.
 
Anyone have any experience here, or ideas for tools 
(besides  packet capture) that can assist ?
 
TIA
 
Ed Maurer

Reply via email to