Re: Consensus on identifier/locator split?

["J. Noel Chiappa" <jnc@ginger.lcs.mit.edu>]

    > From: "Tony Li" <Tony.Li@procket.com>

    > Do we have consensus about splitting the address into locators and
    > identifiers? Note that I'm NOT asking about specifics, like "how big",
    > "what mappings exist", "is it secure", etc. Do we agree that we want to
    > go down this path?

Tony, I'm a little unclear on what your initial question means. Do you mean:
"splitting the 128-bit IPv6 address into two fields, N bits of locator and
128-N bits of identifier", or do you mean "split the concept of address into
two separate concepts, locator and identifier"?


Assuming the latter, I would like to re-make to everyone a point I tried to
make previously, which is that *if you do certain things, you #are#
separating location and identification*, whether you call it that or not.

For instance, if you want to i) keep TCP basically unchanged, and ii) be able
to keep a TCP connection open while switching from one address to another,
then you are, de facto, separating location and identification.

(And as I keep pointing out, MIPv6 is another place that, de facto, separates
location and identification.)


So really the question should not be "do we want to separate location and
identification", which in the IETF always turns into a big wrangle.

Rather, a more productive question would be "do we want to support any
capabilities that require separation of location and identification, such as
keeping TCP connections open when switching from provider A to provider B".

	Noel