Hi all!
(First of all I must apologise for my bad english, I'll try to do my best ;)
Since I visited a seminar about TSM a few months ago I really must say
I'm hooked on the concept and how it works.
I have made a few tests with TSM and have made a decition to 'go sharp'
on one small site that I administer (to begin with) and need some advice.
first, the enviroment:
2 novellservers (one 5 and one 6, most there for testing and deployment,
approx 80gb data, wevy little changes (a few mb/week)
2 domino R6 (one 6 and one 6.5) , I'll integrate these later on as they
have their own backuphandling for the moment, but advice are welcome (I
have'nt read much about this yet, so all I know so far is that I need to
enable transaction on'em).
Changes about: 200mb/day
4 Domino R5 (don't know best practice to backup these)
Changes: 500mb-2gb/day
about 15 Linuxservers (mostly RH4-9, some fedora, and a few RH ES2.1)
changes: 20-50Gb/day
2 of these servers stands for about 90% of the changes (mailservers with
a lot of traffic and access)
10 workstations (W2k and wXP) , very little changes (actually I'm
planning to not do backups on these, as the users is'nt allowed to save
data locally, but you know how it is, when the good looking secretary
comes with her blinking brown eyes it's nice to help her restore the
good recipe on that stew she accidently deleted, and perhaps get a bite
of it too (the stew I mean ;=)
and finally I have about 30 external(running on different type of
xDSLlines , I'm hooked on 20Mbps line) firewalls/servers(linux) with
very little changes (approx 1-50mb/day)
Now over to my question: How do I best plan this enviroment?
The server that I'll be running tsm on is a Linuxserver with 2Gb Ram,
800Gb disk and one DLT)
I'm thinking of having 3 stgpools,
one for backups to land on,
one to let the data queue on for a day or so to not write out changed
files (mostly from the mailservers that has a lot of rewritten files)
and to queue over weekends (when the tape-changer-lady is out-of-office)
and one dltpool
I have runned some tests for a while , and it seems to work okey, only
thing is that I had to set up a seq-diskpool for the dltpool to make
reclamation too, but still there's a lot of tapechanges that has to be
done when a reclamation is running.
How do I trigger reclamations to not use a lot of tapechanges witch are
no good when I only have one drive?
When all abowe servers are migrated to TSM (withing a few months
hopefully, depending on fundraising budgets for the licenses ;) I'll
have an extra backupserver (same config - 2gb ram/1tb disk, one DLT),
I'm planning to make that an backup/failover TSMserver, what's best
practice for that?
move the DLTdrive to the other tsm to obtain 2 drives and use the extra
one to recieve data and send to server #1?
any advices are welcome, specially regarding manual library/reclamation
Kind regards,
Mats
|
