John,
I agree completely! However, we ran into a problem with this plan when
we came up with it. Our servers are located in a glass house where
there are no operators, and we had to have our stackers located in
another building on campus. To do this requires another machine to be
used as a backup server located with the operations staff. When we
decided to do this, we needed the big machine to get the constant
throughput on the DLT drives, thus a SPARCcenter 1000 with two to four
processors to drive the DLTs. I know it costs a bunch of money, but it
was a requirement.
The other issue is the amount of disk on the server. We would still
need to split our backups once the machine has about 100GBs of disk on
it, so we can get under 24hrs. I think we are reaching the maximum
bandwidth on the DLTs now, and we would need a faster drive (IBM?) to
reduce our backup time, or divide the backups into two parts per server.
I think the last issue is the complication to the scripts to do all of
this accross each server...Mic
