| Previous | Table of Contents | Next |
Most Token-Rings chug away pretty well. The most problematic Token-Rings are usually the largest ones, where error recovery doesn’t work as well. (Although you can theoretically put more than 100 workstations on a single Token-Ring, it’s suicidal to go above 60, and in this day of cheap switching, you might as well put one or two dozen per segment.) Nonetheless, Token-Rings are known to go down.
A downed ring usually results from one of these two causes:
A beacon, most often, is when a signal loss occurs on the wire. That is, a workstation has not exited the ring gracefully, and the hub thinks the station is still in use. Because the workstation doesn’t respond anymore (caused by a broken NIC), its downstream neighbor detects a signal loss and tells everyone. (It can still talk downstream, even though its upstream data has been cut off.)
The active monitor initiates a ring purge (or do over). If this doesn’t work, the downstream neighbor thinks to itself, “What if it’s me?” and hurries to check this out. It removes itself from the ring, performs self-diagnostics, and reinserts itself. This cycle keeps happening until the signal loss is corrected. It’s a pretty bad scene until you come along.
Fortunately, you have your trusty Token-Ring monitor program, and the defective workstation has been reported to the active monitor as the reporting station’s NAUN. You walk the NAUN list, find the bad workstation, remove it from the hub, and everything is okay. You can then figure out why the signal loss was occurring at your leisure (it’s usually a bad card or cable). Most times, this is how a beacon troubleshooting session goes.
Excessive errors are troubleshot in pretty much the same way. You can look at the NAUN of the offending machine and remove it from the network. Even though I’ve seen all sorts of errors caused by bad drivers, your trouble will most often be with a single card, cable, or hub. If you don’t have a Token-Ring monitor (and again, some vendors provide this for free-it’s part of what Token-Ring offers over Ethernet) or network analyzer (or if your analyzer provides no clues), the divide-and-conquer method will usually win over any problem.
If you like Macs and BetaMax, you’ll love Token-Ring. Technically superior to Ethernet, Token-Ring nonetheless plays second fiddle to Ethernet on the corporate desktop. It’s able to run over STP and UTP, and with a current ring speed of 16Mb, Token-Ring’s token-passing scheme for wire-sharing means that more people can share the wire efficiently.
Token-Ring’s error-recovery facilities are pretty good, and when these don’t work, the NAUN information that the Token-Ring overhead provides can make it easier for a troubleshooter to find a problem workstation.
Q Seriously, which do you like better? Token-Ring or Ethernet?
A Sorry, I plead the fifth. They both work. Different people like using them for different things.
Q Can I plug a workstation into the ring-in or ring-out port if I don’t have to expand a ring?
A No. Sorry, they weren’t designed for that.
Q Why do I hear an audible click on my older Token-Ring MAUs?
A A physical solenoid is being triggered to extend the ring. Cool, huh?
| Previous | Table of Contents | Next |