Re: COTS Control

On Tue, 14 Dec 2004 11:40:33 -0500, Paula_Stankus_at_doh.state.fl.us <Paula_Stankus_at_doh.state.fl.us> wrote:
> Make sure that both databases are exactly the same. Release to test and
> make sure business users are doing adequate testing. Run the same
> release in the same way to production. If the databases are exactly the
> same then it should work.

Cheers Paula,

actually ensuring that the databases are the same is (relatively) easy - and a reason why I want to spring RMAN10 on my colleagues. Its the various patch levels of libraries/executables;environment settings; ad-hoc reports etc that I'm really concerned about. Stuff that has versions but isn't designed with source control in mind.

A while back I moved a 3 tier app from test to production, this was a 36 stage process and I screwed up one of the stages. Helpfully the new features that the release introduced weren't actually used for 3 months, when they came to be used we got small but significant errors that didn't occur in test. OK so we picked it up after about a week, but the question that was put to me was "how can we audit the live and or test systems so we *know* that they are the same". In an environment where we have source and source control this is doable. I didn't, and don't, have a good answer. Now we did learn some lessons - like don't go live 3 months before you need to and expect a multi-stage process to be implanted in people's minds - but I ought to be able to say, yes this app has the same db, the same executables and the same environment as test. It *will* perform the same functions in the same way, unless testing was inadequate/unimaginative. If I test something i ought to be sure that the live implementation will be the same - and be able to show that it was (or wasn't :( ).

BTW the *databases* were the same - some com+ components were the incorrect version.

-- 
Niall Litchfield
Oracle DBA
http://www.niall.litchfield.dial.pipex.com
--
http://www.freelists.org/webpage/oracle-l