Help to config netbackup parameters. Netbackup 7.5.0.4

[sombramtb]

Hi everyone,

 the structure of servers that we have is this. One master server and 23 Media Servers spread over different towns. All backups starts at the same time, and one of them takes for 350 Gb about 8 or 9 hours (i think too much time), the log is this one.

08/17/15 21:00:00 - Info nbjm(pid=5100) starting backup job (jobid=31347) for client m11edfs011, policy Backup_m11edfs011, schedule Diario
08/17/15 21:00:00 - Info nbjm(pid=5100) requesting STANDARD_RESOURCE resources from RB for backup job (jobid=31347, request id:{91D0E556-7254-4193-9A82-78FED
6C7EE8C})
08/17/15 21:00:00 - requesting resource m11edfs011-hcart3
08/17/15 21:00:00 - requesting resource t11edms011.NBU_CLIENT.MAXJOBS.m11edfs011
08/17/15 21:00:00 - requesting resource t11edms011.NBU_POLICY.MAXJOBS.Backup_m11edfs011
08/17/15 21:00:00 - granted resource t11edms011.NBU_CLIENT.MAXJOBS.m11edfs011
08/17/15 21:00:00 - granted resource t11edms011.NBU_POLICY.MAXJOBS.Backup_m11edfs011
08/17/15 21:00:00 - granted resource 01D016
08/17/15 21:00:00 - granted resource IBM.ULTRIUM-TD3.001
08/17/15 21:00:00 - granted resource m11edfs011-hcart3
08/17/15 21:00:00 - estimated 353441203 kbytes needed
08/17/15 21:00:00 - Info nbjm(pid=5100) started backup (backupid=m11edfs011_1439838000) job for client m11edfs011, policy Backup_m11edfs011, schedule Diario
on storage unit m11edfs011-hcart3
08/17/15 21:00:02 - Info bpbrm(pid=26360) m11edfs011 is the host to backup data from
08/17/15 21:00:02 - Info bpbrm(pid=26360) reading file list from client
08/17/15 21:00:03 - Info bpbrm(pid=26360) starting bpbkar on client
08/17/15 21:00:03 - Info bpbkar(pid=26394) Backup started
08/17/15 21:00:03 - Info bpbrm(pid=26360) bptm pid: 26395
08/17/15 21:00:04 - Info bptm(pid=26395) start
08/17/15 21:00:04 - Info bptm(pid=26395) using 65536 data buffer size
08/17/15 21:00:04 - Info bptm(pid=26395) using 30 data buffers
08/17/15 21:00:04 - Info bptm(pid=26395) start backup
08/17/15 21:00:04 - Info bptm(pid=26395) Waiting for mount of media id 01D016 (copy 1) on server m11edfs011.
08/17/15 21:00:06 - Info bptm(pid=26395) media id 01D016 mounted on drive index 0, drivepath /dev/nst0, drivename IBM.ULTRIUM-TD3.001, copy 1
08/17/15 21:00:02 - started process bpbrm (26360)
08/17/15 21:00:02 - connecting
08/17/15 21:00:03 - connected; connect time: 000:00:01
08/17/15 21:00:04 - mounting 01D016
08/17/15 21:00:06 - mounted; mount time: 000:00:02
08/17/15 21:00:06 - positioning 01D016 to file 2
08/17/15 21:01:29 - positioned; position time: 000:01:23
08/17/15 21:01:29 - begin writing
08/18/15 05:42:40 - Info bptm(pid=26395) waited for full buffer 1073004 times, delayed 1908361 times
08/18/15 05:42:40 - Info bpbkar(pid=26394) bpbkar waited 81941 times for empty buffer, delayed 237382 times
08/18/15 05:42:48 - Info bptm(pid=26395) EXITING with status 0 <----------
08/18/15 05:42:48 - Info bpbrm(pid=26360) validating image for client m11edfs011
08/18/15 05:42:49 - Info bpbkar(pid=26394) done. status: 0: the requested operation was successfully completed
08/18/15 05:42:49 - end writing; write time: 008:41:20
 

[RiaanBadenhorst · Accepted solution]

Hi,

You didn't say if you're taking the backup local to the media server or over the WAN. Assuming its local, increasing the SIZE_DATA_BUFFERS to 262144 COULD improve performance. Please note this is not a given, performance is dependent on many things, data source, data transport (LAN/WAN/SAN) and finally the target.

INSTALL_PATH\Veritas\NetBackup\db\config\SIZE_DATA_BUFFERS

If the file does not exist, create it and add the value above and test the backup. You could also test with 131072 and 524288

[SaurabhHeda]

Hi,

You can also refer the below technote which provides detail information on tuning buffers and improving netbackup performance.

How to configure buffers for NetBackup in a Windows environment to improve performance

https://support.symantec.com/en_US/article.TECH18422.html

[inn_kam]

You can use seeding process also 

          Here are overall steps to finish this task.

1. Restore data for client to a Local machine.

2. Backup this data to Netbackup using one new policy.

3. Configure seeding using this new policy & take backup of Client over WAN.

This will cause to find that client's data in MSDP storage cache. Hence Client will not send exisiting Data over WAN & first backup will finish faster. It will take low bandwidth.

4. Once first backup is successful then remove seeding settings .

NetBackup deduplication client WAN backup: how to seed the fingerprint cache to speed up the initial backup

Article URL http://www.symantec.com/docs/TECH144437

[RiaanBadenhorst]

He's using tape.....

[inn_kam]

ops i have not seen

[Marianne]

" bptm(pid=26395) waited for full buffer 1073004 times, delayed 1908361 times" This means one of two things : 1) slow read speed by client from local disk 2) Slow network between client and media server. Firstly ensure that client backs up to a local media server and not to a remote media server.