Brian E Carpenter writes:
> No, it's the whole point: the classifier at a domain ingress must
> choose an appropriate DSCP value to write into the TC field, and that
> *requires* the classifier to interpret the semantics of some set of
> fields in the packet header.
I don't think this is quite right: you MAY set up a
classifier at the edge of your domain to remark packets,
but it is not a requirement. You could simply believe
that the packet is classified appropriately and police
to the appropriate SLA and remark to the appropriate
outbound traffic class. This allows for the obvious
situation where I, on my own network, have an SLA
with my upstream provider for x% EF traffic, and that
it's completely my own business which traffic I chose
to mark as EF. The upstream provider shouldn't be
either required or even desired to remark the traffic;
that's my problem. His job is only to police to our
SLA.
What you seem to want is a provider-centric
model where you always know better than I do
which traffic should be separated into which
buckets. That's *a* model, but certainly not
the only model, and indeed as is painfully
obvious here, suffers greatly from the
middlebox syndrome as encyption vividly demonstrates.
Mike
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
