No, routing and addressing do not, in and of themselves, require an
identifier in every packet.
I had not understood your question as being focussed on the "in every
packet" question.
I was trying to get at the quesiton of whether the routing and network
system should have an identifier. I think it should. As far as I can
tell, multiple aspects of the system would be cleaner and easier to work
with if such an identifier exists (see below for what I would identify.)
As for "in every packet", no, of course the identifier does not have to
appear in every packet. In some systems, it naturally does. In others,
it doesn't.
I have no problem, for example, with an approach where stateful
communication (such as TCP) sends the identifier in those packets which
are related to establishing or updating the relationship (I.e. the TCP
syn exchange and anything which updates the locator set), but otherwise
uses other information in the packet (a locator plus carefully managed
connection identifiers may be suitable in some approaches) to make
things function. (I am not sure how to handle things like UDP without
putting the identifier in every packet.)
As for what I want to identify, I tend to think that the identifier
should correspond not to the physical node, but to the logical stack
construct. A logical set of protocol machinery, stacked from the
network layer up. A physical machine may have multiple of these. And
in some cases, a single one of these may be operating across physical
boundaries, if the coupling works right.
While I have considered writing up a document on just this, I didn't
bother because it does not seem it would provide any actual value to the
research group in reaching a conclusion on what to recommend.
Yours,
Joel M. Halpern
Scott Brim wrote:
...
Joel M. Halpern allegedly wrote on 12/03/2009 10:57 AM:
My own reaction to this question is similar.
I can imagine an architecture and set of tools where the responsibility
for the identification was either dynamic or an upper layer responsibility.
However, I think that we may be better off, in trying to get to a better
state, if the network / routing / lower layers provides a useful stable
identifier.
Joel, you're talking about an identifier and identification functions
that can be supported in the network layer, so it's a whole-node
identifier. We've been going over this for about a year or more and I
still argue from practicality. A useful stable identifier for what?
Which functions would actually use it?
...
We can dig up old mail if we need to and go through this all again. My
main point in this thread was that routing and addressing, in
themselves, do not require a particular identifier to be in every packet.
Scott
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
