Re: Minimizing backup induced downtime

From: Alexander Skwar <alexander_at_skwar.name>
Date: Fri, 13 Jul 2007 12:40:56 +0200
Message-ID: <1654055.kI3R8puxyT@kn.gn.rtr.message-center.info>

sybrandb <sybrandb_at_gmail.com> wrote:
> On Jul 13, 8:50 am, Alexander Skwar <alexan..._at_skwar.name> wrote:

> Obviously your assertion

"Obviously"... Yeah, right.

>> Let me quote SB: "ALL TRANSACTIONS AFTER YOUR SO-CALLED 'BACKUP'
>> WILL BE LOST FOREVER."
>>
>> So we now finally got to the point, that we can agree that even
>> when using RMAN his statement holds true.
>>

> is just UTTER RUBBISH.

We'll see.

> What's more: YOU KNOW IT.

Nope. Especially since you already said so yourself in <news:gl8d93hv2kn8h4ss81pu19b0uqc6lgh0s4_at_4ax.com>. Why are we now going a step back?

> If you run a database in archivelog mode, use RMAN to backup and
> recover both database and archivelogs,
> you won't loose ANYTHING.

Aha. Just once more, so that I understand it: I finish backup at 22:30. Everything is on tape. At 22:45, something is added to the database. At 23:00, the server dies. I don't have any traces about what was added between 22:30 and 23:00.

From where does RMAN pull the needed data to do a PITR for 22:45?

Remember: Archived redo logs from 22:30 - 23:00 are *not* available. A standby database also does not exist - or it went to flames at 23:00 as well.

TIA, Alexander Skwar Received on Fri Jul 13 2007 - 05:40:56 CDT