> So my question to the forum is have others deployed VTLs in a 
> similar fashion (i.e. MPX = 0) and how have things scaled and 
> is there any associated management headaches? The couple 
> con's of the above approach in my mind would be: 
> * Slower responding NBU GUI because there will be so many 
> devices it will have to manage/query
> * More BPTM processes since each tape drive in use requires 
> one an additional BPTM process
> * Slower backups on a per stream basis (which is OK because 
> you have many drives I guess)

Good assessment.  

I'd stay away from more than 128 drives on a robot, and maybe 128 on a
media server or even 128 on a master.  An admin testing an evaluation
VTL created *396* drives and NetBackup's device discovery found every
one of them.  

- It was annoying to keep track of them (CLI or GUI).
- No point in having 396 drives without tapes for all of them, so that
meant dividing the VTL disk space into very small tapes.
- It seemed (I didn't check stats) to take much longer for NetBackup to
allocate and mount one drive/tape for every job than it did to allocate
a tenth as many drives and multiplex.
- Those little-bitty tapes?  Whether it's the VTL or NetBackup,
something made the frequent dismount/mount activity (because the virtual
tapes were "short") take a significant fraction of what the physical
implementation of drive/media would have taken.
- Neither the GUI nor the tpconfig CLI wanted to work with drives over
index 128.  IIRC, I had to use the tpconfig character-mode menu system
to delete the drives over 128 (or 256, don't remember), and that means
typing the names in for each by-index drive deletion.  Major PITA.

Finally wound up creating something like 32 virtual drives, much larger
virtual tapes, and using heavy multiplexing.  Saw total throughput
increasing all the way from mux=1 through mux=32, though the per-stream
at mux=32 was about a tenth of the mux=1 value.  (I don't want to
publish the numbers as the test scenario wasn't very "real world" and
there was an NDA involved.)  I never tested mux restores in the short
time I had after un-screwing the 396-drives fiasco.

Personally, I'm no fan of VTLs.  Seems like a high price to pay for the
dubious value of having disks pretend to be tape drives. 


_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu