Re: IPv6: what is required of lower layers?

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Wed, 5 Apr 2023 15:45:33 -0400 (EDT)
    From:        Mouse <mouse%Rodents-Montreal.ORG@localhost>
    Message-ID:  <202304051945.PAA18442%Stone.Rodents-Montreal.ORG@localhost>

  | The RFCs I've been using as references for this - 4291 and 4861 - come
  | from 2006 and 2007.  That's not _that_ long ago.

The basic design was all around 1986/87 - everything since then has been 
frills.

  | Complexity can't be hidden, only pushed around.  You have to specify
  | _some_ sort of medium-dependent something, or there's no way to get
  | packets moving.  You can specify a medium-dependent unicast resolution
  | method, or you can specify a higher-layer use of multicast to implement
  | unicast resolution plus a medium-dependent multicast resolution method.
  | Someone must have lost sight of the overall complexity when thinking
  | about the complexity of one small piece.

No, you need to specify a medium dependent multicast resolution method
anyway, as multicast is used for much more than just ND (etc).   The
question is whether you also need to specify a medium dependent address
resolution layer, or whether you just use IP multicast (which is already
going to exist) and something above that.   Whether the "something" is
ICMPv6 or a new Xv6 (layered directly on IPv6, or even on UDP) matters
much less, but I certainly understand the use of ICMPv6 - overall ICMP
isn't used for all that much, yet still has to be implemented everywhere.
Getting more out of it, rather than inventing something new, and similar,
seems like mostly a win.

[Disclosure: I was there when all this was being designed, if I had thought
any of it was the wrong way, I would have objected at the time ... a few
of my comments were even taken seriously, the major addition I suggested
was not.   I still think that was a mistake, but it is way too late now.]

kre

ps: when you're analysing some of this, do remember that the principle
designer of IPv6 (Steve Deering) was earlier also the principle designer
of IP multicast, and draw some conclusions as to impact that might have had.