Hello all!

Because our Tree/network has grown hodge-podge over the years, I wish
to make an effort at better integration and reliability of services.
The current project is to cluster a groupwise service for failover, and
integrate it with an existing tree, displacing the existing GW server
in the process. I understand the failover concept, and have
successfully built a clustered webserver (not in tree or edirectory)
using heartbeat and rsync to provide seamless transition. This method
was great, in that the servers could be placed in separate buildings,
on separate LAN's (or even different towns bandwidth permitting) and
still work.

Where I get confused is with Novell's approach to drive space
management for clusters. All the documentation I have found so far
assumes that there will be a single-source, shared-disk subsystem.
Understandably, such a system would likely have multiple protection
features, but to my mind still represents a single-point of failure
that makes clustering moot. I am looking for my cluster to essentially
be RAID1 servers so if I lose a building to fire or flood, my
organisation does not lose Groupwise services. I have read through the
Novell documentation on installing OES linux & clustering, but I am
still uncertain of the best route.

I ask for patience from the experts here, as I have a number of
questions, though I won't ask them all at once.

Our G/W server is supporting about 300 users, with all of them using at
least a few calendar features, & document sharing. This will not change
more than 10% in the foreseeable future.

Q-1 Can clustering be configured to act more like rsync so that data is
replicated to local disks on each node, rather than a SDS?

Q-2 Can the the above file system use the native linux partitions, or
is NSS the superior choice here?

Q-3 AIUI, NSS has the advantage of being expanded dynamically so that
storage may be added, and this storage does not have to all exist on
the same node- correct?

Q-4 Some posts here have suggested that Novell Clustering may not be
the best choice for Groupwise, but that Heartbeat2 would be better,
especially for the failover I desire. Could such a cluster be
integrated with eDirectory so that the familiar admin tools (ie C1)

Thanks to all who take the time to read & respond