>Has anyone ever heard that rule that lends towards stability of a Novell Server is:
>
>Always make sure that the server processor is equal to or faster than the fastest cpu of all connected workstations.

That's IMO not a very good rule; in fact, I'd say that my Novell servers are probably among the slowest machines on the network in house from the standpoint of the CPU. Their I/O systems can blow about anything else in house off the face of the network, and all of them have LOTS of memory, but its rare to see a NetWare file server becoming CPU-bound.

FWIW, we have a dual P90, and a pair of PII/300 boxes running as NetWare servers, with fairly heavy I/O loading on a Fast Ethernet (100Mbit 100BaseTX), and the other machines on the network range from a low of a Pentium 200MMX up through a PIII/500. We use Adaptec dual-channel UW or RAIDPort adapters on them, and 3COM NICs, and all have at least 128MB of ECC RAM. We've not lost more than a couple of hours due to server failure (wish I could say the same about programmer stupidity) in the past year - in most cases, where we've had a server drop off line, it's just been a matter of switching operating to another server, or in most cases, to replacing one of several disks in a disk array or mirrored disk set. We haven't had a server die for no apparent reason for quite a while.

I'd say that using a stable hardware platform with ECC memory, fast, reliable and fault-tolerant disk storage, and reliable NICs on the servers, and UPS systems on every machine on the network, contribute far more than a faster processor to the stability of a network installation, Novell or otherwise. I'd give very heavy odds that, unless you're running very heavily CPU-bound server applications, the CPU isn't causing the network to flake out...