In article <[log in to unmask]>,
[log in to unmask] (aaron t porter) wrote:
 
>Clayton L. Hines ([log in to unmask]) wrote:
 
>: I plan on finishing up a TACACS-checking process either today or
>: tomorrow.  Once that's done, we'll keep logs of response times from the
>: NDS servers, and set alarms for when the NDS servers don't respond quickly
>: enough.
 
>        What kind of hardware is used and how distributed is this task?
>One would think that this is a rather critical point in the operation of
>the system.  MU-Net isn't worth much if you can't log in reliably.
 
I'll tell you to the best of my knowledge, limited as it is.  We're using
three Novell NDS servers.  I believe that they're dual-Pentium machines,
from some well-recognized vendor.  There's one on each of the Oxford,
Middletown, and Hamilton campuses.  They're each running a TACACS NLM, and
they're configured for some kind of redundancy.
 
I believe that it's just the TACACS NLM that hasn't been completely
reliable in the history of MUnet.  I haven't heard, anyway, of any major
problems with the NDS servers themselves, although Boyd is really the
person to talk on this subject.
 
Clayton L. Hines                                         [log in to unmask]
Unix Software Specialist                  Miami University, Oxford, OH, USA