Can and should may be two different things here.
Can you do what you say? Yes. Each node runs
independently of the others in terms of applications. The degree of
difficulty is way up because you have to design with the idea that a node can
run all or any mixture of apps depending on failover. That brings up
compatibility issues that have to be taken into account. I'm assuming
you're considering a 3x4 where three are active and one is passive to handle the
failover, so the chances could be worthwhile that you'll never have to worry
about running the SQL DB, the jobs, and the query instance all on the same
server at the same time (or a variation of that). A situation could exist that
would however allow all those apps to failover to the single machine at the same
time and that might be a reason not to cluster.
Maybe you could get what you're after with a SQL cluster (2
machine) and then separate machines or even a 2 node cluster for the other
functions?
From: Chris Flesher [mailto:[EMAIL PROTECTED]
Sent: Wednesday, November 12, 2003 11:27 AM
To: [EMAIL PROTECTED]
Subject: [ActiveDir] Cluster service in 2003
I have a question
about how clustering works in general, but specifically 2003. We are wanting to
have a 4 node cluster. However, what I'm not sure of is can you have three
machines running different functions/apps on them have failover capability to
another machine with the "cluster"? There would be a SQL DB machine, a seperate
jobs machine that runs jobs related to the SQL DB machine, and a query machine
that would have point-in-time info from SQL in order to run reports. I want to
know if this configuration can exist in a cluster together and have failover
capability to one machine in the cluster???
Sorry for the
confusing email. As always, Thank you in advance. You guys are a great resource.
Chris Flesher
The University of Chicago
NSIT/DCS
1-773-834-8477