02-05-2014 12:45 AM
Hello
Hope someone could advise – thanks in advance!
I have recently moved all backups to use disk ready for Auto Image Replication to a remote site, however I now have a large amount of jobs queuing. I like to know what you're thoughts would be on our current set up and what settings you would recommend for the storage unit and disk pool? I have reduced the 'Maximum concurrent Jobs' setting from 17 to 15 I have also been toggling with the disk pool Limit I/O Steams setting from unchecked to 3 which made performance worse! I have now increased this to 20, but I am unsure what would be the best setting! All polices backup to disk including incremental changes. The weekly and monthly full backups are only duplicated to tape using Storage Life Cycle Policies.
Our setup is as follows
Version NetBackup 7.6.0.1 All our Clients backup via a dedicated backup LAN 1Gb and full backups are duplicated from disk to tape via fibre.
2 X Media servers with their own 'Media Server Deduplication Pool' - identical setup on each- Windows 2008 R2 Enterprise Edition
Number of policies: 205
Number of Clients: 356 Mixed clients - Windows 2003 - 2008 / Solaris 8 -10 and LINUX
Storage Unit
Maximum concurrent Jobs = 15
Maximum fragment size = 51200
Disk_Pool
Limit I/O Steams = 20
Disk Configuration
Media Sercer Deduplication Pool 1 - EMC SAN Disk
14+1 RAID 5 = 1 x 25 TB LUN SATA (Media Server 1) --- Which at the Windows Level is 50 TB Concat
14+1 RAID 5 = 1 x 25 TB LUN SATA ( Media Server 1) --- Which at the Windows Level is 50 TB Concat
Media Sercer Deduplication Pool 2 - EMC SAN Disk
14+1 RAID 5 = 1 x 25 TB LUN SATA (Media Server 2) --- Which at the Windows Level is 50 TB Concat
14+1 RAID 5 = 1 x 25 TB LUN SATA ( Media Server 2) --- Which at the Windows Level is 50 TB Concat
Solved! Go to Solution.
02-05-2014 06:42 AM
There is no 'best setting'.
The following will probably result in total active streams of 15:
Maximum concurrent Jobs = 15
Limit I/O Steams = 20
Performance Tuning is a lengthy process.
Start with both settings set to 20.
Check that bptm log folders exist on both media servers.
EXamine logs to see what buffer sizes/numbers are used and then at the end of the backups what the end result was - Kbytes/sec and waited for full/empty buffers.
Please bear in mind that 1 Gb network can at best receive data at +- 100 Mb/sec which will limit any hope to see decent performance.
After noting performance with these settings for a couple of days, change settings to 25.
Monitor and record performance.
Making changes in 'baby steps' and recording throughput after each change will enable you to find optimum performance settings for your environment.
02-05-2014 06:00 AM
I Hope someone can advise? Thanks
02-05-2014 06:42 AM
There is no 'best setting'.
The following will probably result in total active streams of 15:
Maximum concurrent Jobs = 15
Limit I/O Steams = 20
Performance Tuning is a lengthy process.
Start with both settings set to 20.
Check that bptm log folders exist on both media servers.
EXamine logs to see what buffer sizes/numbers are used and then at the end of the backups what the end result was - Kbytes/sec and waited for full/empty buffers.
Please bear in mind that 1 Gb network can at best receive data at +- 100 Mb/sec which will limit any hope to see decent performance.
After noting performance with these settings for a couple of days, change settings to 25.
Monitor and record performance.
Making changes in 'baby steps' and recording throughput after each change will enable you to find optimum performance settings for your environment.
02-05-2014 06:58 AM
Hi Marianne
Yes the backup network is the main bottle neck we have.
Thank you for your recommendations. I will follow your instructions in baby steps and see what performance improvements I can find! – I now have a better understanding as to where to look!
Thanks again!