That depends on how fast the filesystems are.

Since you're limited by the 100Mb, you'll probably only end up with a
couple filesystems running at one time.  You can keep one tape drive
busy, or two drives partially busy.0

The number of filesystems you run at once (once you get over one or
two) is probably not going to make any difference on how long it takes
to run the backup.  A 100Mb NIC only runs around 8MB/s, so that's all
you're going to get no matter what you try to do on either end.

Raghavan, Shyam writes:
> Hello Group,
> 
> Does anybody know what is the optimal value for concurrent number of  backup
> jobs a client can safely handle for a typical Unix client with:
> 
> Solaris 2.6
> 4 CPUs
> 2GB memory
> 100 Mbps dedicated hme interface (for backup only) in 100/Full duplex mode
> Total 600 GB of data in  32 Filesystems 
> 
> What is the best way to backup this data using  8 DLT 7000 drives in 2 STUs
> ( 4 drives in each STU)
> 
> Option 1.
> 
> -- Set -max_jobs to 32 for this client
> -- Set MPX per DRIVE to 4 
> 
> Option 2
> 
> -- Set -max_jobs to 16  for this client
> -- Set MPX per DRIVE to 2
> 
> 
> If I use option # 1, then, all 32 Streams from 32 Filesystems will be active
> simultaneously and  I may be really hitting the network interface/ Drives /
> Shared memory pool on client and there by degrading the performance; I am
> not sure ?? 
> 
> If I use option # 2, then only 16 jobs will be active at a time and so it
> may take longer for the backup to finish.
> 
> What do you guys think ? option #1,  #2 or a better 3rd option ?
> 
> Any suggestion or recommendation are greatly appreciated.
> 
> Thanks,
> Shyam
> 
> 
> 
> 
> 
> 
> _______________________________________________
> Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu

-- 
Larry Kingery 
          What has four legs and an arm?  A happy pit bull.