I created policies based somewhat on the function of the groups of
servers I wish to backup.
Most of our windows servers can be successfully protected by a policy
with the ALL_LOCAL_DRIVES file list, and some selective exclude lists on
clients.
Our UNIX systems can be protected by declaring / on the file list, and
then on each host, I create an exclude list. Since the unix systems
are typically database servers, the files and locations of files don't
change much. It is more manual to maintain exclude lists for three
different policies on each host, but it doesn't change very often.
We have NAS servers which operate on their own policy, becasue they take
so long to backup. I start them after the bulk of our unix/windows FULL
backups run, and let them have backup resources as the full backups are
finishing.
We have Exchange backups, both mailbox and DB, which run before the
full/incrementals.
I also try to keep the policies grouped by systems of similar function.
If I needed to do maintenance or change a policy for a specific function
of our environment, I have a few choices. I could disable the whole
backup schedule, or disable just one policy that pertains to the
server(s) that need work. If our unix servers need full cold backups,
instead of the usual schedule, I can create a new and temporary policy,
disable the current one, and avoid changing the policy I use frequently.
We had some horrible SAN issues last year, all related to our unix
hosts. To keep backups running, on the windows side, I simply disabled
the unix policies that were scheduled to run during our maintenance
window. When we finished, I enabled the policy, and did an manual
backup. We occasionally have maintenance on Exchange, and disabling the
mailbox or DB policy for one night is really easy.
I have three policies for each grouping of server. Daily-Incremental,
Saturday-Full, and Sunday-Full. We keep the Sunday tapes on site, and
send Saturday off site.
The Saturday policy is calendar based, and the first Saturday of each
month is set to write to the ARCHIVE pool, with infinite retention.
I would agree with Wayne Smith, in that documenting the policies, and
how they all fit together, is somewhat cumbersome in NB. A method I've
used in the past to document backups is a spreadsheet-looking chart that
shows the hours, and the blocks of time others might expect the backups
to run. It is a visual that even non-technical people can understand,
and it helps me know when the schedule is full, or when people might
expect the system to be finished with backup operations.
-Jon Bousselot
|