Re: [RAM] Ramblings about "locator"

    > From: Brian E Carpenter <brian.e.carpenter@gmail.com>

    > I was thinking about why we're having trouble being crisp about
    > "locator".
    > Maybe part of it is that in some contexts .. a locator is loaded with
    > topological significance.
    > ...
    > Another thought is that on a classical broadcast Ethernet, an Ethernet
    > address is a locator .. it's mapped to topology by the switching or
    > bridging mechanism.
    > Maybe the essential point is that a locator can at least in principle
    > be mapped to topology and an identifier can't.

Sorry, this is confused. You are confusing two (maybe three) different
properties (see my message from yesterday):

- What is being named (an interface, etc)
- What level the name is at (internetwork, or in the case of the Ethernet
	address, network)
- Location sensitivity (i.e. some information about where the thing is is
	encoded *directly* into the name)

"Locator" means different things to different people (for instance, in GSE it
names a subnetwork, not an interface), but AFAIK almost everyone agrees it
has the last property.

Here's a simple test to see if something has this property: can you look at
two *different* names from that namespace, and tell *purely from the names*
whether they are in some way 'close' to each other, in locational terms, or
not? If not, if you have no idea whether the names are close or far, then the
names are *not* locators.

Ethernet addresses say *nothing* about *where* an interface is, you can't
tell from looking at two what their locational relational status is, so they
can't be locators.

Also, be careful with the term/concept "mapping". You can (in some designs)
map an 'endpoint identifier' a la JNC (which is absolutely not
location-sensitive) into some other name, the latter giving you its location.
However, that doesn't mean that because the 'endpoint identifier' can "in
principle be mapped to topology" that it's a locator.

A locator contains, *directly encoded into the name*, information about the
location is - and the 'are these two close' test is a simple test for it.

	Noel

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