Re: management paranoia about 2 million rows updated in 1 commit
From: Niall Litchfield <niall.litchfield_at_gmail.com>
Date: Wed, 22 Aug 2012 20:00:21 +0100
Message-ID: <CABe10saF-i7hoCovQx6S8yvY2Y_tVUFtTubhN1eYNNXBJ2WY7A_at_mail.gmail.com>
I think there are a few things I'd bear in mind. 1. It's not paranoia if they really are out to get you. Bulk updates can really screw you. Very few DBAs know how much undo and redo they will generate for any given statement. Or how long it will take. Answer those questions reliably and you go a long way to getting buy in.
Date: Wed, 22 Aug 2012 20:00:21 +0100
Message-ID: <CABe10saF-i7hoCovQx6S8yvY2Y_tVUFtTubhN1eYNNXBJ2WY7A_at_mail.gmail.com>
I think there are a few things I'd bear in mind. 1. It's not paranoia if they really are out to get you. Bulk updates can really screw you. Very few DBAs know how much undo and redo they will generate for any given statement. Or how long it will take. Answer those questions reliably and you go a long way to getting buy in.
2. It's always worth finding out what happened previously, or the actual cause of their objections.
3. An atomic transaction succeeds or fails 100%. Sometimes 90% success beats 100% failure. And sometimes it doesn't.
4. What resources, how long and what impact can be answered in a representative non production system.
If you can answer authoritatively with it will take this long, have this impact and here's the evidence then most managers will be fine with that. They want reasoned, authoritative argument with accurate risk and impact assessment. So give it to them, after all if you can't its just one blanket statement vs another which already has traction.
-- http://www.freelists.org/webpage/oracle-lReceived on Wed Aug 22 2012 - 14:00:21 CDT