Re: hot backup & direct load - compatible ?

Hot backups are never consistent, unless there is no update, insert, delete activity in the database. When a tablespace is placed into backup mode, the SCN in the header of each datafile which makes up the tablespace is frozen. It is frozen with the SCN of the checkpoint which copied all the dirty buffers which belong to the files which make up the tablespace when the tablespace was placed into hot backup mode. But the datafiles are not frozen. SCNs are placed in the beginning and end of each block written out to the datafiles during subsequent checkpoints. During hot backup, REDO writes complete blocks out to the redo files for those datafiles in hot backup. During recovery, Oracle synchs up the blocks for which the SCNS do not line up. How can they not line up? Oracle cannot control the backup process. The operating system might start copying in the middle of a block to the end of the file, start at the beginning of the file, copy down to the middle of that same block again. Therefore, Oracle writes complete blocks out to redo, which becomes archivelog, so it can fix broken blocks.

To avoid performance problems, schedule hot backup when there is as little update activity on the tablespaces being backed up. Different tablespaces can be backed up at different times.

kstingel_at_sairgroup.com wrote:
: Hi all,

: I have a warehouse which has long sqlloader jobs running against it in direct
: mode. As it happens my hot backup window of 8hrs also overlaps some of the
: direct loader jobs.
: Question - Is my database backup consistent?

: thanks
: Kelvin Stingel
: Atraxis

: -----------== Posted via Deja News, The Discussion Network ==----------
: http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own

-- Received on Fri Nov 13 1998 - 14:49:28 CST