We are installing a standby database for DR side. One of our database is about 10TB and generating daily 500-700Gb archive logs.
Taking tape backups, duplicating them, sending to another city and then finishing restore takes 3-3.5 days long around.
in that period 1.5 TB archive had been generated. Sending these archive logs or taking incremental backup and send via ftp a little bit headache.
So we decided to take clone copy (we said BCV) of that database, take compresses full backup set, ftp to another city and restore.
On production side:
SQL> alter database begin backup
On clone side:
/SYMCLI/V7.1.2/bin/symclone -sid 78 -f /oracle/BCV/xxxpre0_BCV_XXX_devices recreate -noprompt
/SYMCLI/V7.1.2/bin/symclone -sid 78 -f /oracle/BCV/xxxpre0_BCV_XXX_devices activate -noprompt
On production side:
SQL> alter database end backup
On clone side:
SQL> startup mount;
So we can start taking backup.
[xxx:yyyy]@/oracle/dba/rman$ cat rman.cold.DB.full.sh
#!/usr/local/bin/bash
# TODO: Before running this script
# sqlplus /nolog
# conn / as sysdba
# shutdown immediate
# startup mount
# alter system set large_pool_size=1G;
export ORACLE_HOME=/oracle/product/11.2.0/db
export NLS_LANG=AMERICAN_AMERICA.WE8ISO8859P9
export ORACLE_SID=DB
export NLS_DATE_FORMAT=”DD-MM-YYYY HH24:MI:SS”
rman target=/ nocatalog cmdfile=rman.cold.DB.full.rman log=rman.cold.DB.full.log
[xxx:yyyy]@/oracle/dba/rman$ cat rman.cold.DB.full.rman
configure controlfile autobackup on;
show all;
run{
allocate channel c1 type disk format ‘/fra/rman_backup/db_%p_%s_%T.bkp’;
allocate channel c2 type disk format ‘/fra/rman_backup/db_%p_%s_%T.bkp’;
#64 channel opened here
backup as compressed backupset database SECTION SIZE=48G TAG ‘DB_FULL_20120218’;
backup current controlfile TAG ‘DB_FULL_20120218_CTL’;
backup spfile TAG ‘DB_FULL_20120218_SPF’;
release channel c1;
release channel c2;
#64 channel released here
};
exit;
so backup was finished about 7 hours with 64 channel and 100% cpu usage with total size about 1.5TB
So we can ftp the backup files to DR side with 10 parallel ftp session taking about 8 hours.
Now we can start to restore that backup set for standby. we got initDB.ora parameter from prod and we did some changes on it.
SQL>startup nomount
[[email protected]:DB]/oracle/dba/rman$ vi rman.restore.DB.rman
run
{
allocate auxiliary channel c1 type disk ;
allocate auxiliary channel c2 type disk ;
#64 channel was used
duplicate target database for standby
backup location=’/orabackup1/db_full/’
pfile=/oracle/product/11.2.0/db/dbs/initDB.ora
nofilenamecheck;
release channel c1;
release channel c2;
#64 channel was released.
};
exit
Restore operation had finished about 6 hours. So totally 23 hours standby DB had been ready for applying archives;
SQL>alter database recover managed standby database disconnect from session;
So our stand by database was ready and sync with production about 6 hours later.