Re: online backup

I have the same book. The reason the checkpoint SCN is not updated in the file header(s) for a tablespace being backed up is because no updates have been logged to the file since the backup started. The contents of the file(s) does not reflect the current checkpoint SCN. Changes to the tablespace are held in the redo logs until the alter tablespace end backup command is given. At that point changes from the redo logs are applied to the data files and when this is complete the checkpoint SCN in the file header is updated to be current. I would imagine that this is exactly the same mechanism used during database/tablespace recovery. This is why the database cannot be shutdown normal or immediate while a hot backup is in progress, and why Oracle recommends doing online backups when there isn't much activity and backing up only one tablespace at a time (to cut down on redo log usage).

poohland_at_hotmail.com wrote:

> Direct quote from Oracle Backup And Recovery Handbook 7.3 Edition from Oracle
> Press: "Note that when a tablespace is in hot backup mode, only some data
> structures in the data file header (or headers, if multiple dat a files exist
> for that tablespace) are updated, while the others are frozen. However, the
> contents of the file are current. For example, if an update is done on a
> table in this data file, the update is not blocked. But when a checkpoint is
> done, the checkpointed at SCN value is not written to the file header. This
> means that if a crash occurs during a hot backup, Oracle really needs to
> update only the file header and not the contents of the data file while
> opening the database.."
>
> I guess the only thing that got frozen is the file header, but not the data.
>
> Winnie Liu
>
> In article <35D32BD3.5823_at_pharma.novartis.com>,
> Devinder Pal Singh <Devinder_Pal.Singh_at_pharma.novartis.com> wrote:
> > Marcus Reichardt wrote:
> > >
> > > Oracle stops writing to the data files, as otherwise a backup wouldn't
> > > succeed, but it continues to write to redo logs.
> > >
> > > Marcus Reichardt
> > > Consultant
> > There is reaally a confusion there. At one place it says that all the
> > changes to that datafile or tablespace in backup mode goto redo logs
> > that is why more redo is generated. If some data in a block is chanaged
> > then instead of that change going to the redo the whole block goes to
> > redo. The question is does oracle stops writing to datafiles in backup
> > mode and if it does then how these changes are refelected in the
> > datafile and the other part is if it keeps on writing to the datafile in
> > backup mode then that file can go out of sync with the backed up file.
> > All thoughts are welcome
> > Devinder
> >
>
> -----== Posted via Deja News, The Leader in Internet Discussion ==-----
> http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum
Received on Fri Aug 14 1998 - 15:42:45 CDT