You didn't say what clients you are using, but from the "clues" in your note,
I would guess that it is some flavor of Unix. That probably doesn't have
much of an impact here, as scheduled backups run pretty much the same, no
matter what the platform is.
Before we begin, let me go over a couple of basics of scheduling - if you
already know this, Skip on ahead. There are 2 types of schedules - polling
and prompted. All access methods support polling method; only some TCP/IP
implementations support prompted - check your client manuals to be sure. The
difference between the two are sort of subtle. In prompted mode:
1. The client contacts the server saying he wants to schedule a backup.
2. The server acknowledges this, and the client disconnects.
3. When the window for the schedule opens, the server will contact the
client, and start the backup.
In polling mode:
1. The client contacts the server saying he wants to schedule a backup.
2. The server assigns a random time for the client to come back, and
then disconnects from the client.
3. AT the assigned time, the client establishes a session and does the
backup.
The significant difference here is who starts the scheduled session. When
polling is used, backups occur randomly through the first half of the window,
while with prompted, the server will begin doing the backups as soon as the
window opens. Multiple scheduled sessions can be started at the same time,
but if there are more than ADSM can handle, he will wait until one is
finished before starting another.
I would assume that since polling is the default, and in many cases the only
option, I would assume most people will use it. We have people doing both,
but most use polling - I prefer the randomization.
Now, how to set up the schedules....
Schedules are all set on the server, so there is central control of them. A
node is assigned to a schedule (or schedules - but let's leave that for
scheduling 102). He cannot choose which one he will use.
Start the schedule by entering the DSMC Schedule command. It varies between
platform, but the recommendation that I make is to use whatever startup
facilities your platform supports to have the schedule start when you boot
the machine.
Remember that to ADSM, the first Incremental backup is by default a full
backup. I prefer to have these run via schedules in the middle of the night.
When the schedule is set up by the administrator, the default is to do an
incremental. Most schedules I think are set up like this, with no overriding
options or parms set. If you follow the defaults, you will get a safe,
incremental process that should handle most situations. I only do special
things (like selective backups) when I need to give special consideration to
something such as an SQL server or DB. These also are topics for Scheduling
102 :-)
SO - scheduling is really pretty straight forward:
1. Assign a node to a schedule
2. Start the schedule
3. Go home and enjoy the evening - don't bother with two aspirin
(unless you have a headache from something else) :-)
4. Check it in the morning!
And feel free to contact me if you have any other questions. Happy ADSMing.
Jerry Lawson
jlawson AT thehartford DOT com
______________________________ Forward Header __________________________________
Subject: basic scheduling 101
Author: INTERNET.OWNERAD at SNADGATE
Date: 1/20/98 2:58 PM
hi all,
I am a "new" user and I am attempting schedules for the first time.
What are the minimal pieces I need?I have set up the following
dsm.opt schedmode prompt
dsm.sys schedmode prompt
defined a schedule with the objects I hope to selectively backup and
associated that schedule with a client node
set schedmod to any on the server
issued a dsmc schedule on the client
Is this the simplest solution? I hope to control as much as possible thru
the server to hopefully "centralize" my schedules? Any information is
gratefully accepted.
thanks , Ann Courchaine
Year 2000 National Practice : Consulting & Systems Integration
|
