wsmith@umn-cs.cs.umn.edu (Warren Smith [Randy]) (08/30/88)
Since the IEEE 802.4 standard has been mentioned several times in this group recently I was wondering if anyone has actually used the priority class mechanism defined in the standard. For those who aren't familiar with the standard there are basically 4 priority classes. Each station is allowed to send a fixed amount of data in the highest priority class each time it receives the token. In all the lower priority classes stations may only send data if the token is rotating around the ring quickly enough. If the token is rotating quickly enough that a station can send packets in a low priority class then that station may send packets in that priority class until it has slowed token rotation time below the threshold for that priority class. (for the sake of brevity this description is a bit simplified) If a majority of the traffic on the network is being sent in one of the lower priority classes and the network is heavily loaded service becomes quite irregular (though bounded within the limits described in previous notes). Things may actually degenerate to the point where one station fills the entire alloted bandwidth for a single rotation, then the token passes completely around the ring (with no stations able to send data in this particular priority class) and to its succeeding station which may proceed to do the exact same thing. To those who have used 802.4 or participated in its creation: How is this priority mechanism intended to be used? Does anybody actually use it? Does anyone really tune their station management parameters? -- Randy Smith wsmith@umn-csw.cs.umn.edu ...!rutgers!umn-cs!wsmith
eli@spdcc.COM (Steve Elias) (08/30/88)
In <7083@umn-cs.cs.umn.edu> wsmith@umn-cs.UUCP (Warren Smith [Randy]) writes: > >To those who have used 802.4 or participated in its creation: > Does anyone really tune their station management parameters? a definite yes! there are .4 parameters that need to be tuned depending on the backbone size and the response time you want. it is possible to trade off throughput against response time, to some extent -- by varying the 'token hold' time. i believe that it is possible to thrash up a token bus network if you add a station that is improperly configured. steve elias, Chipcom
morgan@Jessica.stanford.edu (RL "Bob" Morgan) (08/31/88)
Steve Elias writes: > i believe that it is possible to thrash up a .4 net if you add a > station that is improperly configured. Gee, I hope I don't have to use one of these nets in *my* nuclear power plant . . . - RL "Bob"