[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RAM] Ramblings about "locator"



Noel,

On 2007-06-15 14:50, Noel Chiappa wrote:
    > From: Brian E Carpenter <brian.e.carpenter at gmail.com>

    > My original RAMble (sorry) ended thus:

    >  Maybe the essential point is that a locator can at least in
    >  principle be mapped to topology and an identifier can't.

My human-name can be mapped to a telephone number. My human-name is not,
however, a telephone number. An "identifier" can certainly be mapped to a
network location with the aid of the appropriate database, just as a DNS
name can be mapped to a network location.

Maybe you're using the wrong word, and instead of "mapped" you really meant
something like 'directly related', or something like that, but the problem
is that that's a somewhat nebulous turn of phrase. Which is why I
previously suggested that nice gold-standard, easily-comprehendable tests
for 'does this name have location information embedded in it' are:

Well, let me quote Schoch:

>>     At the time one wishes to communicate with a particular address, there
>>     will be some mechanism that will map an address into an appropriate
>>     route.
>>
>>     The address (where something is) need not be bound to the route (how to
>>     get there) until this mapping takes place; the choice of an "appropriate"
>>     route may change over time.

The thing is, Schoch didn't separate the loc and id aspects of addresses
(after all, the whole point was a single logical addressing scheme for the
whole catenet). So this issue we're discussing didn't arise. However, what
he writes seems to me to apply to locators.


- Do you have to get a new one when you move to a new location? - Given two location-based names, can you tell, by comparing *just the names*, whether they are 'close' to each other?

Maybe someone else can come up with a snappy word/phrase which clearly,
cleanly and concisely embodies that very close relationship...


> A slightly different way to say it is that a locator is a handle for > a route.

No. My street address is not a handle (in any simple sense) for directions to
my house. Yes, with a lot of data (connectivity information), and a certain
amount of computing, it can be turned into directions, but the relationship is
not a simple one.

True, but inside a RIB, a locator is definitely going to be what you use to find a route.

Brian

This was all quite clear a long time ago; see IEN-19, John Shoch, "Inter-Network Naming, Addressing, and Routing", January 1978, now finally available at:

    ftp://ftp.isi.edu/in-notes/ien/ien19.txt

Everyone here should be familiar with what it has to say.

	Noel


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