Thank you John for your 'archeology' work about this protocol number. I appreciate your time and effort.
About the draft itself, I will let the authors reply (as they should be part of int-area@) but my understanding is that the "multi-access gateway" is on the same link (3G or wifi) as the client so it is a layer-3 0-hop. And IMHO hop-limit should be 0 to scope the packets to the local-link (even if less secure than using HL=255). Regards -éric On 20/09/2019, 02:02, "John Gilmore" <[email protected]> wrote: Protocol 114 was unassigned in RFC 1700 in Oct 1994, which was the last RFC tabulating protocol assignments. In January 2002, RFCs ceased being published for protocol number assignments, according to RFC 3232. Sometime before Feb 1999, protocol 114 was assigned here: https://web.archive.org/web/19990203044112/http://www.isi.edu/in-notes/iana/assignments/protocol-numbers The original IANA, Jon Postel, died on October 16, 1998. There was some turmoil in the relevant websites at the time. The Internet Archive's Wayback Machine does not appear to have captured the IANA.org or isi.edu websites during an earlier time when this protocol number was not assigned. But, only five assignments in Feb 1999 had followed 114; the next one was L2TP (protocol 115) by Bernard Aboba (April 1998). The preceding one was PGM (protocol 113) by Tony Speakman in January 1998. So it's a pretty good bet that it was assigned by Postel between January and April 1998. (L2TP was documented in RFC 2661 of August 1999, and by that point it was not using protocol #115; it ran over IP and UDP on port 1701. A later 2005 evolution of L2TP, L2TPv3, used protocol 115.) Does anyone have archives of the TCP-IP Distribution List from 1998? The only copy I have found so far is at http://securitydigest.org/tcp-ip/ but it ends in 1994 (with no apparent "we're closing down the list" messages). A separate issue: Having read the draft-zhu-intarea-gma-03.txt, and skimmed the 2017 draft-kanugovi-intarea-mams-protocol-03 that it references, I don't see how this protocol could in any way be seen as a 0-hop protocol. The whole design is to provide multiple paths to the Internet, which would require that the relevant packets traverse routers. The MAMS draft explicitly says "MAMS routes user plane data packets at the IP layer". 0-hop protocols only operate on a single LAN and cannot be routed, by definition. (ARP, DHCP or its predecessor BOOTP are examples of 0-hop protocols.) Therefore, I think this draft should not be using protocol 114. John _______________________________________________ Int-area mailing list [email protected] https://www.ietf.org/mailman/listinfo/int-area
