In article <1D06C9567A6AD4119391000629D9009D9760C8 AT central.ssd.jhuapl DOT 
edu>,
"Prather, Wanda" <Wanda.Prather AT JHUAPL DOT EDU> says:
>
>Any body out there doing some backups over Frame Relay that would be willing
>to answer a few questions?
>We have a customer that is planning to do it with some NT clients and is a
>little nervous.
>
>Are you doing it?

Yes
>Approx. how many clients?

65

>About how much data per day?

30-50gig

>
>Any major problems other than the obvious issue that speed is slower than
>for local backups?

Performance issues both in legnth in time to do the backup and conflicts
with other network use. Lately the network folks have changed something
which makes this worst. But we are going to start using the new block level
backup option to see if that helps.

Maybe Andy or Mike can tell us if TSM will look at single instance storage
across all clients on a TSM server and maybe QOS or pacing for help with
these problems.

One problem that was a surprize is the expire inventory process. This
was normally scheduled to run on sat mornings, but at this time we normally
have some long running client sessions. With the latest algorithims used by
the expire inventory process, the log file can be locked during the expire
inventory run and after, until all the long running clients come to a commit
point. With some of the clients with the right files this can be several
hours. Therefore this leads to more log filling. We used to have an adsm
vm server with over 1000 clients. The expire inventory on the server would
run for 6 days and we did not fill the log which was less then a third the
size. But then the expire inventory does seem to run faster now. Much faster.

PS I am guessing at the exact details of the locking on the log, but it seems
   to fit what I see, and the TSM folks have not published a how-does-it-
   work book for me to look it up. Other then what slips out here.

len