Joel M. Halpern allegedly wrote on 02/01/2010 11:41 EST:
> As far as I can tell, the MP-TCP thing works without having such an
> end-point identifier. Their view, as I understand it, is that they
> establish communication based on some location pair. They then use that
> communication to establish enough shared sessions tate that they can
> relate other communication streams (TCP connections) to the initial one,
> to form the full session.
> Thus, the location independent identifier is a dynamic value determined
> once communication is established.
Yes, although I wouldn't call them TCP connections.
> It seems to me that having a stable long term location independent name
> for the communicating entities would be helpful, but I have not done a
> good job of articulating why.
>
> Yours,
> Joel
>
> PS: While I agree with the points I elided, I do wonder if we really
> need a different term for the location insensitive, stable name when we
> use a short binary string vs when we use a thing that looks like a DNS
> name.
I never named them, but here's my list. I think you're talking about #2
and #3.
* identifiers for access, to use a visited network at all.
* identifiers for initially finding something you want to talk to.
Examples would be domain names and SIP URIs.
* identifiers used for initial contact, in order to establish
sessions.
* identifiers used for session control: initial authentication and
association of locators with sessions, as well as
re-authentication when locators change.
* identifiers used in referrals, whereby one endpoint tells another
about yet a third.
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
