Re: [tsv-dir] FDDI operation (was RE: tsv-dir review of draft-ietf-vrrp-unified-spec-02.txt)

> AFAICT, this could cause a packet to circulate forever; FDDI doesn't
> have a counter AFAIR (it was a *long* time ago...)

The problem is not in the normal token passing process, but the process for 
electing a master, which is responsible for generating the clock for the 
entire ring, keeping a bit FIFO to absorb jitter introduced by phase locked 
loop hunting, and other management functions such as re-injecting tokens after 
a hiccup as well as several self diagnostic functions.  AFAICR the mac address 
in used for tie breaking when a new master is elected.  From the words in the 
vrrp draft, I have to assume that duplicate mac addresses cause the algorithm 
to fail.

In the normal token passing process the station which injects the frame is the 
one to removes it.  This is not done by matching the addresses, but by 
steering the bits in the layer 1 serialization logic.  Normally the sending 
station is removing the head of the frame while it is still sending the tail. 
A given station can send multiple frames, but at some point it must send a 
special "token" frame.  Any station which has traffic to send can claim the 
token (by flipping one symbol) thereby changing it into a a packet preamble, 
and the process continues with a new sender.

Loosing the token (a ring of all idle symbols) is detected and corrected by 
the master.

Yea, we had lots of FDDI at one point.

Thanks,
--MM--
-------------------------------------------
Matt Mathis     http://staff.psc.edu/mathis
Work:412.268.3319    Home/Cell:412.654.7529
-------------------------------------------
Evil is defined by mortals who think they know
"The Truth" and use force to apply it to others.
_______________________________________________
vrrp mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/vrrp