Buffer_Size parameter still needed when using direct connect to appliances?

[Beavisrulz]

Hello all,

We've always added the following registry key to our clients after installing NetBackup:

Buffer_Size=128  (decimal value)

This was originally started many moons ago when we were having issues with backing up to tape. Now, 90% of our backups go directly to our disk appliances, using client-side dedupe and Accelerator. The backups that we have to run to tape for long term retention do of course use our Windows 2008 media servers.

My question is: do we still need this setting on most or any of our clients?

Our master server and media servers are Windows 2008R2, and we have 3 NetBackup 5220 Appliances. The registry value has just always been a part of our client install process and we didn't know since our environment has changed drastically since we first implemented this setting back in 2009, if we still needed it or not. I'm also just wondering if this setting is causing our backups to run slower than what they could be running, based on reading the following tech doc:

http://www.symantec.com/business/support/index?page=content&id=TECH28339

We also still have old config files from 2008 on our master server. We left them there after we migrated from 6.5 to 7.1, and the operating system from Win2003 to Win2008R2.

SIZE_DATA_BUFFERS = 262144

NUMBER_DATA_BUFFERS = 32

Our backups that go to tape are running pretty fast, so I don't want to make any changes that will adversly affect the performance of those backups. Please let me know what (if anything) should be done with these config settings, or if we can eliminate them all together.

Thank you for your input!

Larry

[Marianne · Accepted solution]

The TN has been updated recently saying that the NBU tuning should be disabled (buffer value set to 0).

.... disable tuning by placing a zero (0) into the NET_BUFFER_SZ file on media servers and UNIX/Linux clients.  Simply deleting the file is not equivalent because some NetBackup process have default setsockopt API calls configured to overcome past external problems with various platforms and drivers.

$ echo '0' > /usr/openv/netbackup/NET_BUFFER_SZ

On Windows clients, the same effect can be obtained by placing a 0 into this registry key.

HKEY_LOCAL_MACHINE\SOFTWARE\VERITAS\NetBackup\CurrentVersion\Config\Buffer_Size

....

To determine if manual tuning can provide better performance than auto-tuning; record the current O/S settings, then temporarily adjust the TCP memory allocations upwards in increments of 65536 or 131072 bytes and observe if the changes are beneficial.  Also ensure that TCP window scaling is enabled.

So, why not make it 0 on one client as see what the effect is?

As far as buffer settings on the master server is concerned:

SIZE_DATA_BUFFERS = 262144 
NUMBER_DATA_BUFFERS = 32

These settings will only be used if the master is also a media server.
If this is working well for you, leave it as is.
You can view the effect of buffers on backup performance in bptm log on each media server.

Each media server need to have own buffer settings.
They have been pre-added on Appliances.

This Tuning Guide should still be relevant and explains how/what to check in bptm logs:

NetBackup™ 7.0 - 7.1 Backup Planning and Performance Tuning Guide  http://www.symantec.com/docs/DOC4483

Long story short:

If your performance is good, leave settings the way they are.

*** Update ***

 NetBackup Backup Planning and Performance Tuning Guide, Release 7.5 and Release 7.6 :

http://www.symantec.com/docs/DOC7449