Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Minimizing backup induced downtime

Re: Minimizing backup induced downtime

From: joel garry <joel-garry_at_home.com>
Date: Thu, 12 Jul 2007 14:53:43 -0700
Message-ID: <1184277223.516519.250640@o61g2000hsh.googlegroups.com>


On Jul 12, 2:31 pm, Alexander Skwar <use..._at_alexander.skwar.name> wrote:
> · joel garry <joel-ga..._at_home.com>:
>
> > Please see Metalink Note:302615.1 RMAN and Split Mirror Disk Backups
> > for how to keep the downtime very small and use RMAN to backup,
> > keeping the Oracle and non-Oracle parts synced by splitting them both
> > at once. Maybe I still don't get it, but I tried.
>
> I don't have access to Metalink, but I'll see if our Oracle
> support can send me this document tomorrow. Thx.
>
> It shows how to backup non-Oracle parts with RMAN? Interesting.
> I hope they can send me this metalink document.
>
> > Another odd thought, get directories of the non-Oracle files into a
> > file, mount that external and CTAS into the db.
>
> Hm. Interesting thought. How would I make sure, that the directory-gone-file
> is always up to date? I guess this would mean, that the app should
> also update this directory-gone-file when it modifies the files in
> the directory? So I'd have to double the storage - once I need to
> store the files as files in the vault directory and once as a
> directory-gone-file?
>

Well, hopefully a directory listing is smaller than your files :-) That was just an off-the-top suggestion, implementing it is an exersize. But the key to that sort of scenario is diffs, and when to make them.

>>>From where does RMAN/Oracle
>>>pull the data about what has been done between 22:46 and 22:59,
>>>if archive logs aren't available?

>If that's somehow possible, then you really got me hooked on
>RMAN.
You need to read the basic manuals if you don't understand this. When you backup online, the files are in a fuzzy state. When you recover, logs are applied to make them unfuzzy. So with a PITR, you can recover until whenever you want. If the archived logs are not available, it is not possible. That is why you want to back them up, and that is why you want to not lose any after the backup. There are a number of scenarios where you need to restore where you haven't lost the archived logs. For example, you lose a device that has data files on them. One of the data files has never been backed up. You can recover the data file and its data even so! I believe there is an example of this in the manual.

Metalink access is really useful.

jg

--
@home.com is bogus.
Real money from virtualization: http://www.technewsworld.com/rsstory/58241.html
Received on Thu Jul 12 2007 - 16:53:43 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US