The flow label has always been defined as immutable, but for examples that break the standard, see draft-hu-flow-label-cases.
For the rest I agree with Joel. Regards Brian Carpenter On 2010-09-25 14:40, Yiu L. Lee wrote: > Hi Joel, > > Sorry for my ignorance. Can you explain to me what is mutable flow label or > pint me to a reference I can read? You are right, this usage wasn't intended > to substitute for transport protocol and port numbers in ECMP and LAG. But I > guess FL could be used for other purposes other than ECMP and LAG as long as > it wasn't caused any conflict. > > Thanks, > Yiu > > > On 9/24/10 10:25 PM, "Joel M. Halpern" <j...@joelhalpern.com> wrote: > >> It is quite a stretch to claim that all traffic originating from (or in >> the other direction destined to) a single customer constitute a >> meaningful "flow". However, because RFC 3697 was carefully written to >> be vague about this, it would be difficult to prove that it is incompatible. >> >> I would note that this usage of flow label would be inconsistent with >> mutable flow labels, and would be inconsistent with the desire to use >> flow label as a meaningful subsitute for transport protocol and port >> numbers in ECMP and LAG logic. Whether either of those two incompatible >> desires will themselves be standardized is extremely unclear at this >> point, although there seems to be significant resistance to having flow >> labels be mutable. >> >> Yours, >> Joel M. Halpern >> >> On 9/24/2010 10:14 PM, Yiu L. Lee wrote: >>> Hi gents, >>> >>> We have a design question of Flow Label. During the v6 transition, some DSL >>> providers may want to create an IPv4-in-IPv6 tunnel from the BRAS to the >>> AFTR to continue to provider v4 access over a v6 core network. To identify a >>> CPE behind the BRAS, we propose to use the Flow Label. Each CPE will be >>> assigned with a Flow Label. This Flow Label represents a flow of all encap >>> v4-in-v6 traffic behind a CPE. The Flow Label will be applied on the v6 >>> address of the BRAS. v6 hosts behind the CPE will have their v6 addresses >>> and be most probably from a different v6 prefix, so their flow labels won't >>> be affected. >>> >>> You can find the details in: >>> http://tools.ietf.org/html/draft-zhou-softwire-ds-lite-p2p-02 >>> >>> Our question is: "Is this usage compatible to RFC 3697?" We posted this >>> question to Softwires and we were told to also ask 6man for input. >>> >>> Thanks, >>> Yiu >>> >>> -------------------------------------------------------------------- >>> IETF IPv6 working group mailing list >>> ipv6@ietf.org >>> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 >>> -------------------------------------------------------------------- >>> > > -------------------------------------------------------------------- > IETF IPv6 working group mailing list > ipv6@ietf.org > Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 > -------------------------------------------------------------------- > -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
