Different workgroups/networks...

[Paul Dekkers]

Hi

I'll first describe our setup:
We have two different networks, one for teachers, and one for students. They
are divided from each other, except for ip traffic, and it should stay that 
way, because students are not allowed to reach the teacher's server...
(So when using samba more, I'll have to setup a better firewall I think,
and/or samba rules from which IP shares may be mounted...)
Currently both servers are running Novell Netware.

Planned was to use NT Server in the future (for at least the theacher's
server), but since it is easier under UNIX to create accounts (with scripts), 
and we have a UNIX server (already with accounts for each teacher and
student), I wondered if it was possible
to do all authentication via the UNIX server. Is that possible? And is it
still possible to divide the two networks from eachother? 
I thought maybe
it would be nice if users in group @doc just could login on the teacher's
network, and in the group @stud just in the student's network. Is that
possible?
Is that possible with NT servers for as well students and
teachers (so that the primary login is on the students or teacher's server), 
or even with no NT servers at all? (So that both networks also use the
server for their primary domain login...)

BTW, I'm totally new to using the UNIX server as a domain logon server, but
I hope I'll get from it what I expect. Is it e.g. possible to place the
desktop of the user, and his/her settings on the server, so that all
workplaces are "transparant"?

Looking forward to any reactions, and thanks in advance,
Paul

P.S. Maybe it's possible when all users do their domain logon on the UNIX
server, and than in their login-scripts the drives on the NT server's are
mounted? (And that the NT servers are protected with a firewall?)
However, I would like it more if as well students and teachers would login
on their own NT server, and that that NT server get's its authentication
from the UNIX server...

P.S. I hope you understand what I want :-) (I'm afraid it was not such a
clear explanation...)

--
Paul Dekkers 
E-Mail: <P.Dekkers at cgu.nl>
To err is human, to moo bovine