OUI != 00-00-00

[Lawrence J Lang <nvuxe![e-mail]>]

paul>Dave Piscitello raised the question of whether we can expect problems
paul>from using a Protocol_ID of 00-00-00-Ethertype...
Actually, Paul, that was me.  I just Cc'ed to Dave (he works with me,
and I thought he might be interested).

paul>The answer is NO, provided the rules in RFC 1122 are followed.  That RFC
paul>states that nodes may use 802.3 frames per RFC 1042...provided that they
paul>ALSO accept and respond to "traditional" Ethernet frames
paul>...
paul>It was based on that rule that we defined FDDI-Ethernet bridge translation.
paul>...

I was thinking more of the case where the SNAP uses
an OUI that is *not* 00-00-00.
paul>If you don't insist on an EtherType (i.e., you're willing to accept
paul>values other than 00-00-00 in the first 3 bytes of the Protocol_ID) then
paul>there are lots of others that could provide the value, probably sooner.

Then, an FDDI/Ethernet bridge would translate such an FDDI frame into
an 802.3 frame, rather than a traditional Ethernet frame, on the 'orange hose'.
No problem for 1042-capable systems, but it would lock out 894-only systems.
This may not matter if EARP is only of interest to systems on the FDDI ring,
but this effect seems to militate against specifying such a SNAP.

The SNAP should probably use the OUI 00-00-00,
and an EtherType assigned for EARP.

Larry Lang

Bellcore
[e-mail]