Conference 7.286::fddi

2224.0. "MAC Frame-control priority bits" by NSDP01::RV () Tue Feb 18 1997 13:38

Hi,

I was asked to trace the performance of an FDDi ring made of DECBridge6XX,
Wire concentrator 6XX, DECSwitch900 EF, on which Alpha and HP systems are
connected to.

Nothing abnormal was detected and the traffic is low (5 to 10% without ring
purger), but the customer told me that he had remark that starting file tranfer 
between HP to HP and in parallel one of the involved HP with systems on Ethernet 
seems to work with higher priority for what is comming from Ethernet (through
DECbridges or Switches). It is only when the ethernet transfer is finished that
the FDDI (HP to HP) can continue, so he is asking me if there are possible
reasons why our bridge/switches can monopolised the token ?

For what I was able to measure (because it was not possible to simulate these
transfer while I was there), is that HP systems are using when sending frames
Priority equal to 7 while all the other systems (and of course traffic comming
fro the bridge/switches) are using Priority 0. Is it somewhere a reason that
could explain the slow activity of the HP systems.
Or is rather a I/O problem or queuing issue on the HP systems

Remark that without Ethernet acess FDDI transfer between HP is quite good. 

Thanks in advance
Robert Vandenberghe

        See notes 319.* and 1561.* for the pre-existing discussions on the
    FC priority bits.
    
    MDL