Thomas,
reading your article I think that a real randomization is not what you
want to make happen.
When you add a node to the schedule, you can set a start time
individually for each node. Only if more nodes have the same starting
time, the system decides on the backup order. I think ADSM always starts
the next scheduled node even if the system was busy on the scheduled
point in time, when the possible backup window you set for the node is
not yet over.
It took a few optimization steps to find a schedule for our servers. On
one hand you'll want a schedule where you can rely on preset starting
times, on the other hand there should be not idle times for the server.
Just play around and give it some tries to get the right schedule.
Regards,
Michael Bartl
--
Michael H. Bartl, ADAC e.V. (IDL/PPO)
Michael H. Bartl, ADAC e.V. (IDL/PPO)
Tel.: (089) 7676-4063, Fax: (089) 7676-8161
EMail: Michael.Bartl AT Zentrale.ADAC DOT de
>
|