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: RMAN Question

Re: RMAN Question

From: Ryan Gaffuri <rgaffuri_at_cox.net>
Date: 9 Dec 2003 08:45:02 -0800
Message-ID: <1efdad5b.0312090845.16f7c2dd@posting.google.com>


"Howard J. Rogers" <hjr_at_dizwell.com> wrote in message news:<3fd58a4b$0$1011$afc38c87_at_news.optusnet.com.au>...
> "Ryan Gaffuri" <rgaffuri_at_cox.net> wrote in message
> news:1efdad5b.0312081133.75351dfe_at_posting.google.com...
> > >
> > > So I presume the issue is simply what backs up the RMAN repository
> itself?
> > > I'd suggest export/import. The repository database should be small; the
> > > repository information is smaller still (about 20MB per target database
> per
> > > year); and you only need to export the one RMAN schema. Will probably
> take
> > > about 10 minutes tops.
> > >
> > > Regards
> > > HJR
> >
> > you cant detect block corruption with exp/imp. No dbverify for .dmp
> > files.
>
> We're talking about a very small tablespace being exported, not gigabutes of
> stuff. Export is its own check on corruption within the source data file,
> because it's doing selects, and any corrupt blocks will cause it to blow up.
> No you can't check the dmp file for corruption with dbv, but then that's
> what import does.
>
> Alternatively, try transportable tablespace backups. Then you have a dbf,
> and dbv can verify that to its heart's content.
>
> Or alternatively still, do a boring old 'alter tablespace rmants begin
> backup' and copy the damned thing at the O/S level.
>
> Why anyone wants to turn this particular molehill into a mountain, beats me.
>
> >Not ideal. Id consider, but I really think Id go with 2 RMAN
> > repositories or maybe even three to be super anal.
>
> That's just silly, not anal. Two separate repositories requires two separate
> backups of the targets, which is twice the workload, twice the I/O and twice
> the repository management issues. The issue isn't that major in the first
> place, and your approach is utter over-kill, potentially confusing, and
> therefore potentially dangerous.
>
> When all is said and done, RMAN doesn't even need a repository, even before
> 9i.
>
> HJR
i posted on here a few months ago and the consensus was to always use a repository with rman. what is your take?

I forgot about transportable tablespaces. I like that strategy. I can run dbverify on that. Received on Tue Dec 09 2003 - 10:45:02 CST

Original text of this message

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