Pascal et al. -
can you clarify the role of contexts relative to (vs.) the IPv6 flow label?
It seems to me that we are talking about establishing contexts that will
be appropriate for certain kinds of flows.
For most applications today, and certainly many applications in the
future, the overwhelming majority of packets flow multi-point to point.
If I understand what you're doing, this type of flow would likely have a
single context, and there wouldn't be too much overhead to set it up and
maintain it, and I'd be able to do pretty good header compression as a
result of that.
But wouldn't the same be true if I just used a flow label? I think that
I'd get all of the benefits of the existing context proposal, and on top
of that be able to elide almost everything after the flow label.
Flow setup and maintenance should be the same as with contexts (I think?).
Maybe this makes Carsten happier too, since f(flow label) -->
everything, and the setup and maintenance has to explicitly define that
mapping.
ksjp
Pascal Thubert (pthubert) wrote:
Dear lowpanners
Draft -03 says:
How the shared context is assigned and managed is beyond
the scope of this document.
In fact, all HC needs is that the information necessary to expand the
addresses is present in the context table and indexed by information in
the packet (interface ID, source/dest short addresses, context ID or
default).
So HC would work in different environments, for instance:
- only a default context and but a table indexed by short addresses
- multiple contexts but just one prefix per context
- multiple interfaces, one prefix per interface
There's no point in enforcing any of these case in the spec. I wish to
add this text:
The information that is maintained in that shared context is out of
scope.
Resulting in:
2.1.2. Context Identifier Extension
This specification expects that a concept of context is shared
between the node that compresses a packet and the node(s) that need
to expand it. The specification enables a node to use of up to 16
contexts. How the contexts are shared and maintained is out of
scope. What the context information is is out of scope. Actions in
response to unknown and/or invalid contexts are out of scope.
What do you think?
Pascal
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
