Ayuda de vídeo de Screencast

RMAN backup is runnning very slow

Created: 14 Nov 2012 | 6 comments

RMAN backup is runnning very slow, average speed is 1MB/sec. How to find out what is making the backup slow.

Around 100 GB DB is taking 14 hours to complete the backup. It was taking only 4-5 hours to complete the Tape backup before, now we have moved this oracle backup to Data Domain ATMOS storage. Not sure why it is running very slow, Data Domain mean it should complete fast right.

 

Help me to find out what is wrong, that making the backup slow.

Please let me know how to proceed and which are the logs required to bottleneck this issue.

Comentarios ComentariosIr al último comentario

el cuadro de los Marianne

Help us to understand backup parameters in rman script. 
filesperset value? How many channels?

Datadomain recommends filesperset of 1 and multiple channels (sbt_tape).

Supporting Storage Foundation and VCS on Unix and Windows as well as NetBackup on Unix and Windows
Handy NBU Links

el cuadro de los Yasuhisa Ishikawa

On what point did you determine average speed? Average tranfer for each backup jobs? Or by DB size devide backup elapsed time?

I assume you determined by formar way because of extremely low performance. On RMAN stream backup, Oracle does not pass whole database blocks to NetBackup - only used blocks. So backup size will be lower than database size. If only few GB(or MB) is used, average transer seems very low from NetBackup.

Even if only part of database space is used, Oracle read whole database space to check its consistency. So if space usage is low, read performance at oracle side become key factor. You should check disk read performance and busy rate on Oracle host.

Authorized Symantec Consultant(ASC) Data Protection in Tokyo, Japan

el cuadro de los louisborder

For my backup same problem here..( Dead Slow) .could you please me out?

I am a professional blogger, present started writing artilces on job descriptions

el cuadro de los shinnet

       SID    SERIAL#      SOFAR  TOTALWORK Blocks/sec Mins to go START_TIME           LAST_UPDATE_TIME     % COMPLETE

---------- ---------- ---------- ----------             ---------- ----------          -------------------- -------------------- ----------

        16         11    2242104    6102080                146               439              17-nov-2012 18:59:16 17-nov-2012 23:14:07      36.74

        17         23    1582200    8925568               103                1183            17-nov-2012 18:59:19 17-nov-2012 23:14:07      17.73

 

23:20:00    %usr    %sys    %wio   %idle

Average        3          11           1         85

 

load averages:  0.66,  0.60,  0.54;                    up 119+02:57:57                                                     23:19:42

101 processes: 100 sleeping, 1 on cpu

CPU states: 92.3% idle,  2.3% user,  5.2% kernel,  0.2% iowait,  0.0% swap

Memory: 16G phys mem, 7957M free mem, 15G total swap, 15G free swap

el cuadro de los Yasuhisa Ishikawa

Sorry, I have mis-understood as you use ATMOS as primary storage. You used ATMOS as backup destination.

So, first thinfg you should to do is measuring ATMOS performance on your media server. Check activity with tools like iostat, perfmon or so. depending on your OS and how you configure the access(NFS/CIFS/...) to ATMOS. Or do performance test with backup of large local files.

Authorized Symantec Consultant(ASC) Data Protection in Tokyo, Japan

el cuadro de los shinnet

We had issues because all the backups were running at a time and the backup was running slow. We have created seperate storage unit for this RMAN backups and all the backups are running fine now.