Re: [RAM] Re: Ramblings about "locator"

[jnc at mercury.lcs.mit.edu (Noel Chiappa)]

    > From: RJ Atkinson <rja at extremenetworks.com>

    > since the IEEE MAC address is used in a lookup table .. to forward the
    > frame/packet, therefore the MAC address must have location semantics
    > and so could not be a pure Identifier.

Actually, I think I probably said something as follows:

  If we define 'locator' is a 'internetwork-level location-sensitive name for
  an interface', then one obvious design option for a 'full' locator, for the
  typical Ethernet interface, is some sort of hierarchical name for the
  physical network (i.e. subnet), with the MAC address appended.

In other words, an IEEE MAC address is *part* of a locator (if you are
defining 'locator' to be 'internetwork-level location-sensitive name for an
interface').

Yes, if you plug that interface in somewhere else, the locator changes - but
the low-order part (the MAC address) will be the same. Similarly, because to
me a 'locator' is the name of an interface, the MAC address (or something
that maps to it, as in IPv4) would have to be part of the locator.

Does that make sense?


    > I am pretty confused at this point about what your perspective is. I
    > gather your thinking has evolved since this precise question last came
    > up.

No, not really. The things I think have changed are that i) I'm more
sensitive now to understanding how all these terms (e.g. "locator") all mean
different things to different people, and ii) I'm slowly becoming more
precise (with all these different axes of name attributes: locationality,
level, object, etc) about the taxonomy of names.

	Noel

_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram