The IP over ATM Mailing List Archive by date

[Andy Malis <malis@enigma.acton.timeplex.com>]

Andrew,

> I know this is an IETF list, but I'm not that comfortable with the
> multiprotocol implications of choosing an OUI which is under IANA
> control. Does IANA mind administering numbers for things like
> NetBIOS and IPX too?

I have two answers:

1. In the general case, they have been doing so (at least reporting,
   if not in fact registering) for quite a long time now - take a good
   look through RFC 1700 sometime ... it's a fun read!  Really!  (Now
   I know the definition of a Real Geek :-)

2. In this specific case, it doesn't matter one way or the other.  The
   encapsulation format is:

  [0xAA-AA-03] [0x00-00-5E] [New PID] [2octet CMI] [Old PID] [MP Packet]

  The only thing being registered is [New PID] under OUI [0x00-00-5E].
  This identifies a single new protocol, "MARS Cluster Data Transfer".
  Note that the embedded "[Old PID]" uses Ethertypes (also listed in
  RFC 1700) to identify which network-layer protocol is being
  transferred.

> Grenville has very carefully made sure (at least in versions -01
> ... -04; I haven't yet digested -05) that MARS stuff is all very
> protocol-neutral and it would be clumsy if it used 00-00-5E for IP
> and a different one for other protocols.

No problem (see above).

> If the choice of OUI really is arbitrary then I would suggest using
> one that is protocol-neutral e.g. 00-A0-3E (and we could get some
> PIDs next week if time is important :-)

Thanks, but we should probably should be vendor-neutral as well ... :-)

Cheers,
Andy
__________________________________________________________________________
Andrew G. Malis   malis@maelstrom.timeplex.com             +1 508 266-4522
Ascom Nexion      289 Great Rd., Acton MA 01720 USA   FAX: +1 508 266-2300