Understanding Accelerator (from detailed Status view)

[rblaas]

Hi all,

I am in need of confirmation or explanation on some logging I see in the detailed status.

When I see this in the detailed status (for an Incremental backup):

04/15/2015 07:02:43 - Info bpbkar (pid=98149) accelerator sent 1115317760 bytes out of 5630247936 bytes to server, optimization 80.2%

Does this mean the Accelerator has found 5630247936 bytes of new/changed data on the client of which 1115317760 bytes is new data to the backup appliance? 

And how does this related to:

04/15/2015 07:02:45 - Info netbackup (pid=164455) StorageServer=PureDisk:netbackup; Report=PDDO Stats for (netbackup): scanned: 5498334 KB, CR sent: 268206 KB, CR sent over FC: 0 KB, dedup: 95.1%, cache disabled

I see that the client-side-dedup has found scanned: 5498334 KB of new/changed data of which sent: 268206 KB is sent to the Backup Appliance as this is new unique data.

Please correct me where I am wrong or comfirm I am understanding this correctly.

Many thanks

Regards,

Ronald

[sdo]

Yes, that's the way I read it too.  Accelerator found that it didn't have to read/process everything that it discovered.  And de-dupe found that it didn't have to send everything that was read.

Was this for a Windows based client?  If it's a Windows based client, you may be able to achieve further optimisation if you have 'Change Journal' selected/enabled in the Client Configuration.

[rblaas]

This was from a Apple (10.6.8) client. 

Any optimasation tips are welcome!

Thanks for your confirmation.

[sdo]

Ok.  I'm not aware of any further optimisation tips specific for MaxOSX file system backups.

One thing to perhaps consider (and test?) might be to ensure that the client side NetBackup 'network buffer' size is zero - and thus allow the TCP stack in MacOSX to manage its own network buffering:

http://www.symantec.com/docs/TECH28339

[sdo]

Top tip:  Backup is one thing, restore is another.  It's usually a good idea to try to perform a test restore (and maybe a compare/diff too - just to be sure) of new backup clients, and especially for OS variants/flavours which might be 'new' to your environment - and then capture and document the restore process for these new OS variants, and also share this documented process with your local admin/support/tech teams - as it might be some time before you, or someone else, has to perform an urgent restore.

[rblaas]

yes the Buffer Size is set to 0 

And yes Always perform restores (disaster tests)

[watsons]

Yes, I think you read it right:

5630247936 (divided by 1024) is about 5498334 KB.

268206 is about 4.9% of  5498334 , so dedup rate is 95.1%.