Below... On 2010-04-15 21:14, Alexandru Petrescu wrote: > Le 14/04/2010 23:51, Brian E Carpenter a écrit : >> Alexandru, >> >>> I find this proposal to change the Flow Label behaviour to come in >>> too early, at a point where we don't yet have widespread use of >>> simple Flow Labels (or is it widely used?). >> >> But that's the problem. If you have read the -02 draft, you will >> understand that flow label usage today is essentially non-existent. > > Ok, it is draft-carpenter-6man-flow-update-02 and I will comment on it > separately. > >> Usage is not going to occur unless we make it desirable - and it's >> quite clear that RFC 3697 did not succeed in making it desirable. So >> either we do something, or we continue to carry 20 useless bits in >> every IPv6 packet. > > I mainly agree to make Flow Labels desirable. Towards that end one > would consider two other cases where Flow Labels were considered: > > - MEXT' WG draft-ietf-mext-flow-binding-06 "Flow Bindings in Mobile > IPv6 and NEMO Basic Support"; > - ROLL WG's intention of use of Flow Labels in RPL > protocol draft-ietf-roll-rpl-07. > > Currently their way of considering flows and Flow Labels respectively is > incoherent with old Flow Labels - and with new Flow Labels too. Or > probably are they about to get in synch(?). > > Why has MEXT WG defined a new flow id instead of reusing the IPv6 Flow > Label? Is it because of security? The new 6MAN Flow Labels don't > improve on security, they're still uncovered by IPsec. > > Why has ROLL WG looked at mutable Flow Labels? Is it to change it > enroute and restore it when getting out of the flow label domain? The > new flow labels don't seem to impose the restoration, in my reading. > > Are the new Flow Labels better adapted to MEXT and to ROLL?
Well, if we get to a stable proposal here, we cn ask them exactly that. Brian -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
