Hi Brian: I think this went unanswered, sorry about that.
RPL voted not to use the flow label because we were afraid we could "shoot ourselves in the foot" by using a method that could become invalid. Instead, the group elected to go to the RPL option in a Hop by Hop header as was presented to 6MAN. This resolution causes RPL border routers to insert additional bytes in every packet for 1) the RPL header, and 2) the IP in IP encaps that comes with it. That really hurts in our constrained environment. The initial approach (up to draft 7) was to use the flow label as specified in http://tools.ietf.org/html/draft-ietf-roll-rpl-07#section-7.2 . As you see, that approach would also leave 12 bits to the network to network communication, and 8 for the app to network communication, in our case to tag a routing topology with the so-called instanceID. If we had pursued that path, we'd probably have had to further specify the 8 bits, in particular to allow for the case where a source leaves the FL it to 0 and the BR actually sets it, and to indicate the FL scope, that is which section of the network the FL setting is supposed to affect (eg source subnet or destination subnet or global). Cheers, Pascal > -----Original Message----- > From: Brian E Carpenter [mailto:brian.e.carpen...@gmail.com] > Sent: Thursday, July 29, 2010 11:35 PM > To: Michael Richardson > Cc: Pascal Thubert (pthubert); ipv6@ietf.org > Subject: Re: Flow Label: 12 bits mutable and 8 bits immutable > > On 2010-07-30 01:11, Michael Richardson wrote: > ... > > If the ROLL WG does not feel conformtable with the above, then I think > > we need to go ask some grandfathers of IPv6... What's the FLOW LABEL > > FOR then? > > That's the problem. It was never determined. That's why RFC 3697 had such > limited goals. > > The introduction to draft-hu-flow-label-cases-00 gives a short summary of the > history. > > > > > I think our usage of flow label is completely within the pervue of > > what was intended. > > Can you summarise that use for those of us who don't follow ROLL, and indicate > how it plays with the rules in RFC 3697? > > Thanks > Brian -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
