Hello Mike

My observations has been that Netbackup will initially allocate an active tape
for any given backup run if available, such a tape would be partially filled but
once that backup run uses up that tape, it allocates a fresh tape rather than
another active tape in order to continue.   The end result is that the number of
active tapes in a given pool tends to grow to the number of available drives or
the number of concurrent jobs in the class, whichever is lower.   It's very
difficult to get Netbackup to roll down the number of active tapes in a given
pool after having changed the number of drives available or the number of
concurrent jobs allowed other than to suspend the active tapes and wait for them
to expire.

Chances are that the filesystem on the Netapp box is such that the backups
typically fill an active tape and continue onto a fresh tape thereby making the
fresh tape a partially full active tape so the cycle begets itself.

Seems to be a design decision to favour brute performance rather than efficient
tape library management.

The tape allocation from the scratch pool seems to be somewhat bizarre as well.
It's definitely not first in, first out or least recently used as logic would
dictate.   Does any one know if there's any kind of reasonable algorithm being
used such as by oldest last mount date or lowest mount count?   It would
certainly seem to me that no effort has been made to manage the tape library for
equitable media use or to maximize the expired media residency in the scratch
pool as an added protection window, really why does the catalog forget all about
the images on a tape when expired? why not simply note them as potentially
unavailable and forget about them when the tape gets re-assigned, importing
expired tapes would be a lot easier if there was a way to know which tapes to
import (even Microsoft had the forethought to do a recycle bin...).   Then
again, what do you expect of people who use mount count instead of an end to end
pass count?

Regards,
John I Wang
Sr. Systems Engineer
Steverson Information Professionals

---
Enron Broadband Services
Enron Building 1472c
ph (713) 345-4291
fax (713) 646-8063


|--------+----------------------->
|        |          mtraves.news@|
|        |          home.com     |
|        |                       |
|        |          11/17/00     |
|        |          06:46 AM     |
|        |                       |
|--------+----------------------->
  >-----------------------------------------------------------------------|
  |                                                                       |
  |       To:     veritas-bu AT mailman.eng.auburn DOT edu                      
 |
  |       cc:     (bcc: John Wang/Contractor/Enron Communications)        |
  |       Subject:     [Veritas-bu] NDMP on NetApp                        |
  >-----------------------------------------------------------------------|



I installed the NDMP extension for NBU 3.4 a week ago to backup a NetApp box to
direct attach DLT drives.  Schedule was set up to do Full backups once a month,
Differential's weekly, and Incremental's daily.

Although the configuration seems to be working properly, we've noticed that each
day, regardless of the schedule type, a new tape from the Scratch pool is
selected for backup, even though the previous days tape isn't full -- hence the
question:  Is the NetApp box via the NDMP extension capable of "appending" to a
previously written to tape, or should I expect a new tape (or tapes) to be used
every day?

Cheers,

Mike