Netb 6.5.4,Solaris with SAP and ORA agents.
Problem: for scheduled backups, there appear to be some issues related to media selection or 'why did my job stop'.
Let me explain.
With SAP backups I see this: policy starts, single stream, picks a tape,brbackup writes to it (SAP with Oracle btw).
Then the backup enters a second phase, ie a separate job is spawned. This is to do the redo phase,ie we run brarchive.
It drops the first tape and selects another ,writes to it.This repeats for the remaining phases of the job, so theres 2 tapes written alternately. Each spends (wastes?) time dismounting and remounting a new tape,A,B,A,B,A.
Not knowing SAP for I'm not a SAP admin, is theres a good reason for doing this ie not putting all the data on a single media?
I ask since on an unloaded system its not that important but when on a heavily used system this causes an issue: some other job jumps in and grabs the drive...backups that should take no time to complete to do arch redos or utl files (backup runs 5 phases,5 changes of media) are stretched out over hours.
Similarly wth Oracle...the tpl runs, I write out the bulk of my data and then the next phase runs (archivelogs) and it needs a different media.Something else jumps in as if the job hasnt started and my Ora backup hangs around. This gets worse..the next phase (when there is a drive free) is then regarded as being out of the window of opportunity...its done the bulk data and now its been canned because the window has closed.
Surely it would make sense to say if the initial job started in the window then all other spawned jobs are therefore deemed to have started within the window? Is there a way to do this?
Simlarly if the job has begun then tape drive alloc should be reserved for that job if the job or subjob is still running?
Thanks in advance,jim.
