Hi everybody.
Still green on SAP & Oracle so be patient.
Oracle Database 11g Enterprise Edition 11.2.0.3.0 running on RHES 6.3 x86_64.
We perform regular, online backups using RMAN with the following scripts:
## full backup - every 2 weeks ##
RUN
{
CONFIGURE BACKUP OPTIMIZATION ON;
RECOVER COPY OF DATABASE
WITH TAG 'EMT_MMFG_backup'
UNTIL TIME 'SYSDATE - 30';
BACKUP
INCREMENTAL LEVEL = 0
DATABASE PLUS ARCHIVELOG
TAG = 'EMT_MMFG_backup';
sql 'alter system archive log current';
sql "CREATE PFILE=''/oracle/EMT/oraflash/backups/EMT.ora'' FROM SPFILE";
copy current controlfile to '/oracle/EMT/oraflash/backups/EMT_control.ctl';
}
## incremental backup - every day ##
RUN
{
CONFIGURE BACKUP OPTIMIZATION ON;
RECOVER COPY OF DATABASE
WITH TAG 'EMT_MMFG_backup'
UNTIL TIME 'SYSDATE - 30';
BACKUP
INCREMENTAL LEVEL = 1
FOR RECOVER OF COPY WITH TAG 'EMT_MMFG_backup'
DATABASE PLUS ARCHIVELOG;
sql 'alter system archive log current';
sql "CREATE PFILE=''/oracle/EMT/oraflash/backups/EMT.ora'' FROM SPFILE";
copy current controlfile to '/oracle/EMT/oraflash/backups/EMT_control.ctl';
}
RMAN's retention policy is set to "CONFIGURE RETENTION POLICY TO RECOVERY WINDOW OF 7 DAYS".
On our ERP/TEST system we noticed the following behaviour, starting with a clean, empty Flash Recovery Area (located in a separate filesystem):
1) on day one, we performed a full backup
According to a query performed against Oracle, approx 417GB of data was written to the Flash Recovery Area and this was confirmed by an inspection of the area itself:
# du -csh *
131M autobackup
409G backupset
4.0K datafile
2.8G flashback
4.0K onlinelog
412G total
Please notice the backupset which is as large as the DB itself (well, a bit less mainly due to the compression alghoritm being used); the datafile folder is empty.
2) on day two, we performed an incremental backup
According to the query, only 260MB of data was written into the FRA but disk usage changed dramatically:
# du -csh *
261M autobackup
410G backupset
549G datafile
2.8G flashback
4.0K onlinelog
962G total
As you can see, the backupset folder did change a little bit whereas the datafile folder (which was empty after the full backup) got to 549GB in size!
3) on the following days, incremental backup operations did not change the scenario much and still backupset and datafile folders were populated.
When approaching the full backup, we ran out of Flash Recovery Area space and we had to reduce the retention to (eg) 3 days before removing obsolete backupsets (RMAN> DELETE FORCE OBSOLETE;).
After executing the above command, we noticed that the datafiles folders was still at 549GB whereas the backupset one was reduced to 30GB.
My question is: is it normal to have both backupset and datafile folders with such a large amount of data?
I have to admin the size of the EMT Test db is quite large (2/3 times the size of our productive environment, mainly due to a number of client copies executed against it), about 550GB (the size of the datafile folder) and that this behaviour seems to be the same on other TEST/PROD environment but, again, I'd like to know more about RMAN's space allocation policy for the Flash Recovery Area.
Thanks,
Rob