On Jul 1, 2009, Patrick Frejborg wrote:
I think host identity together with a generic session identification
mechanism would make it easier to promote the decoupling, there
would be
more good reasons to upgrade the stack - e.g. server load balancing
and
troubleshooting in the "middleboxed" Internet would be a lot more
easier
to carry out.
Patte -
So you are saying that it would be useful if (part of) a session
identifier identified the host that is keeping the session state.
Perhaps. Or, more specifically, session identifiers should also
identify the /stack/ that is keeping the session state.
But I am still unconvinced that session identifiers really should
identify anything else but a session. There are two reasons: First, if
they did, session mobility would be constrained, because a session would
by definition be bound to something else, such as a host or a stack.
Second, given a pure session identifier and the session's address, I
don't yet see a general benefit of having a host (or stack) identifier
in addition.
The emphasis here is on "general", since there will certainly be
applications that need host or stack identifiers. But again, those host
and stack identifiers will be application-level components, and don't
have to be defined as part of Internet architecture in my personal
opinion.
- Christian
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
