Re: RMAN Restore with Incremental

From: Jason Heinrich <jheinrichdba_at_gmail.com>
Date: Thu, 3 Apr 2008 11:53:59 -0500
Message-ID: <b32e774d0804030953h5a391684sd93b7f8332b43a3@mail.gmail.com>


Charlotte,
You are correct, only the changed blocks will be backed up into the backup set. I believe Robert was referring to the fact that the entire datafile will need to be read by RMAN to determine which blocks had changed, making the backup just as slow as a full backup. Thus the importance of the BCTF to reduce time and I/O.

On Thu, Apr 3, 2008 at 11:32 AM, Charlotte Hammond < charlottejanehammond_at_yahoo.com> wrote:

> Hi Robert,
>
> Can I confirm I understand correctly your remark "one block change in a
> datafile will cause the entire datafile to be backed up" ? If I have a
> (full) 4Gb datafile, say, and change a single block (eg. a single row in a
> table), then the whole 4Gb datafile will be backed up in a subsequent
> incremental level 1 backup?
>
> This doesn't really seem to tie in to what I've seen in the past - my
> understanding was only that changed blocks would go into the incremental -
> regardless of the use of BCTF. Could you point me to the right place in the
> manual - please!
>
> Thank you very much!
> Charlotte
>
>
> ----- Original Message ----
> From: Robert Freeman <robertgfreeman_at_yahoo.com>
> To: Charlotte Hammond <charlottejanehammond_at_yahoo.com>
> Sent: Thursday, April 3, 2008 2:44:38 PM
> Subject: Re: RMAN Restore with Incremental
>
> The smaller backup set size could reflect a change in the underlying data,
> a mass delete of rows for example.
> Without a block change tracking file, one block change in a datafile will
> cause the entire datafile to be backed up
> in an incremental. Hense, the BCTF is pretty darned important! :-)
>
> RF
>
>
>

-- 
Jason Heinrich

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Apr 03 2008 - 11:53:59 CDT

Original text of this message