RMAN script advice(Hot full daily backup)

Question

I am currently running a daily hot full backup (by customers request).
My code is executed as an "at" job nightly, but I am running into some hits on performance of the server at the time my backup is running. Here is the code I run:
rman target /
run {
backup database plus archivelog;
delete noprompt obsolete;
sql 'alter system archive log current';
exit;
My RMAN show all;
retention policy to recovery window of 3 days;
backup optimization off; #default
default device type to disk; #default
controlfile autobackup on;
controlfile autobackup format for device type disk to '/path/%F';
device type disk parallelism 1 backup type backupset; #default
datafile backup copies for device type disk to 1; #default
archivelog backup copies for device type disk to 1; #default
channel device type disk format '/path/%U';
maxsetsize to unlimited; #default
archivelog deletion policy to none; #default
snapshot controlfile name to '$ORACLE_HOME/dbs/snapcf_sid.f'; #default
I do not have a recovery catalog established- customer request also.
I am running the buggy 10.1.0.2.0 on Linux AS3. My normal backups should only last 1 hour 20 minutes, but it varies between 12 hours to 5 hours to 3 hours- never consistent.
Any thoughts?

Answer 1

So CPU performance seems to be inline, but you may want to check I/O during the backup with vmstat.
You can also query V$BACKUP_ASYNC_IO and look at EFFECTIVE_BYTES_PER_SECOND, WAITS and IO_COUNT.
During the backup, you can run the following query to monitor RMAN:
SELECT sid, serial#, context, sofar, totalwork,
     round(sofar/totalwork*100,2) "% Complete"
     FROM v$session_longops
     WHERE opname LIKE 'RMAN%'
     AND opname NOT LIKE '%aggregate%'
     AND totalwork != 0
     AND sofar <> totalwork;If the % Complete doesn't seem to be increasing, you should run the following select to determine what waits exist:
SELECT sid, seconds_in_wait AS sec_wait, event FROM v$session_wait
     WHERE wait_time = 0
     ORDER BY sid;Even though you are using automated SGA, you may want to specify LARGE_POOL_SIZE = 16m.