Re: Oracle direction on triggers

replies embedded . . ..

On Mon, 24 Feb 2003 12:13:03 -0800, DA Morgan <damorgan_at_exesolutions.com> wrote:

>Ed Stevens wrote:
>
>> We have a developing situation I'd like some feedback on.
>>
>> A few months ago a new application went to production, using an Oracle
>> 8.1.7 db on Win2k. This app is partly purchased product, partly
>> custom code. Recently, another app, running on DB2 on the mainframe,
>> needed to have a copy of the data on one of the tables in this system.
>> We settled on using a trigger and the OTG to keep a DB2 table
>> populated with changes from the Oracle table. The lead analyst on the
>> Oracle app got upset over this approach, claiming it would wreck
>> performance. He further stated that "Oracle apps have set a direction
>> of moving away from the use of triggers because of performace." Since
>> I don't keep up with Oracle Apps (we have none here) I couldn't refute
>> it, but it sure sounded specious to me.
>
>Your developer is likely steeped in mythology. Could it hurt performance?
>Of course. Will it? No one knows.
>

Oh, he is most definately steeped in mythology. He keeps saying they have to wait for the trigger to finish its work. Well, of course they do, but how long is that wait? A microsecond? Two seconds? Twenty seconds? They can't say, which leads me to believe they can't really percieve it, meaning the wait is more theoretical (in the microsecond area) than real (in the multiples of seconds area).

>A change like this should be implemented in a test environemnt and tested.
>This is not an appropriate discourse for someone guessing off the top of
>their head.
>

I guess I have to take the hit on that. I tested it for proper operation, but did not benchmark the performance.

>Triggers, in the past, were more of a perforance issue than they are
>today. With well written code I'd certainly consider using triggers a
>viable option at least until an objective test told me otherwise.
>
>Daniel Morgan
Received on Tue Feb 25 2003 - 07:27:48 CST