Re: Does it matter where the binaries are?
- From: stephen booth <stephenbooth.uk@xxxxxxxxx>
- To: adar76@xxxxxxxxxxxx
- Date: Thu, 10 Mar 2005 17:36:49 +0000
On Thu, 10 Mar 2005 19:57:44 +0200, Yechiel Adar <adar76@xxxxxxxxxxxx> wrote:
> I would strongly recommend against it.
>
> Just think, you need to put in a patch. So, you will take ALL the databases
> down, install the patch and then have to upgrade ALL the databases at once.
> Now, suppose you have problem with one of them and you want to go back to
> the old version, you need to go back with all the databases.
The arguement is that if we have the binaries spread accros half a
dozen servers then we'd have to apply the patch half a dozen times.
The plan is that before we applied the patch we'd use NetApps
'snapshot' funtion that provides you with an instant point in time
disk state that you can rollback to in moments. As I mentioned in
another response, this is somethign I got hit with in the middle of a
meeting today, it's still in the concept stage.
>
> Also you need to shut down all the databases in the company. What do you do
> if the accounting can be shut down only from 01:00 to 05:00 and the SAP can
> be shut down only from 19:00 to 22:00?
>
The systems they're talking about this for are actually different
installation of the same application (Documentum from EMC, it's a
document management system). So it would be Planning's docbase,
Housing's docbase &c. I'll advise against it but when it comes down
to it organising the outages isn't my problem, it's the service
deliveries manager's problem and he seemed really interested in this
idea. If he wants to sow the wind, he can reap the whirlwind.
Stephen
--
It's better to ask a silly question than to make a silly assumption.
--
//www.freelists.org/webpage/oracle-l
Other related posts:
