Re: Some one tested RESETLOGS new functionality on 10g?

From: Tim Gorman <tim_at_evdbt.com>
Date: Tue, 01 Mar 2005 16:45:13 -0700
Message-ID: <BE4A4C99.240E0%tim@evdbt.com>

No trick question, just straight-forward DBA stuff...

We have a winner!

Mark, I owe you a tall beer or a short single-malt, whenever we meet...

on 3/1/05 4:38 PM, Mark Bole at makbo_at_pacbell.net wrote:

>>>> POP QUIZ:  The need for a "cold" backup immediately following RESETLOGS was
>>>> one of at least two situations where a "cold" backup was "better" from a
>>>> data protection standpoint than a "hot" backup.  There remains one
>>>> situation
>>>> where a "cold" backup is better (from a data-protection standpoint) than a
>>>> "hot" backup.  Can anyone name that situation?
>>>> 
>>>> To put it another way, what is the situation where a production DBA might
>>>> ask to halt a production application to take a "cold" backup, and not be
>>>> crazy?
>>> 

> [...]

>> Anyway, since your database was closed for the PITR, a "hot" backup isn't
>> really possible there.  My question was intended for the situation where
>> either choice was possible.
>> 
>> Anyone?  Bueller?  Bueller?  Anyone?
>> 

>
> If this is a trick question, then the answer is: when your database is
> in NOARCHIVELOG mode, a cold backup is better from a data protection
> standpoint than a hot backup.
>
> OK, here's a guess: you lose all members of a inactive on-line redo log
> group that has not yet been archived... (at least, that's one scenario
> mentioned in the docs). Similar reasoning would apply after any
> NOLOGGING operation in a database or tablespace without FORCE LOGGING in
> effect.
>
> Although I'm not sure why in any such situation, you couldn't archive
> all the online log groups, and then take a hot (inconsistent) backup?
>
> -Mark Bole

--
http://www.freelists.org/webpage/oracle-l

Received on Tue Mar 01 2005 - 18:47:04 CST