Afaik, a bptm is in charge of getting a backup stream to a tape.  So if 
you have MPX of 10, you'll have 10 backup streams to each tape and 10 
bptm processes.  Just the way it is.  1,000 policies sounds like a lot 
... do you have multiple clients per policy or multiple policies per 
client?  Maybe separating the master from media server(s) is in order?  
Having the scheduler look for work less often does sound like a good 
idea.  Let us know what you find, please!

cheers, wayne

Veritas Netbackup wrote, in part:
> Our netbackup master-media server a SF6900 with 4 dual core CPU's and 
> 16 Gb of memory is 100% utilized all the times. The only time it gets 
> some respite is during the afternoon, when users do not want to load 
> their system with backup processes.
>
> Even with 20 Active backups our server indicates a minimum of 95 % 
> util. I checked ps -ef and found numerous bptm and bpsched processes.
>
> How can we reduce the bptm processes??? I spoke to an ex-colleague and 
> he opined that we change the scheduler invocation to 20 mins from the 
> default 10 mins in the global attributes. Since we have around 1000 
> policies and hence the scheduler might not able to complete its first 
> round and meantime nbu would invoke another one. This overlapping he 
> said shd be reduced.
>
> Any suggestions to reduce CPU load on netbackup server, especially as 
> to what might be invoking these bptm processes??????