You are correct, Carsten: This also relates to the concept that was evoked earlier in the thread to allow an implicit format on a RPL link. There's no such thing as a RPL link. As you say, RPL control itself does not exhibit the HbH header since most of it is 1 hop - link local. And then a link upon which RPL is active can be shared with other activities that are not RPL related at all. If the HbH is present, then the route lookup will be done along a VRF that is indexed by the instanceID. Other tagging might come in the future. And then there's the default RIB when the router has no better indication in the packet.
I like your vlan analogy. It is quite usual in practice to map vlans onto vrfs. Since we are doing route over, we are quite naturally integrating the concepts. For the time being and considering the constraints that apply to RPL, we found that one octet was the right trade off to represent an instance. We'll probably see a day, though, where the concept generalizes, and at that time we'll probably need 4 octets at least. In particular, RPL or something very much like RPL would probably be very interesting in hierarchical access domains like DSL where 1:1 vlans apply today. Take care, Pascal > -----Original Message----- > From: Carsten Bormann [mailto:[email protected]] > Sent: Tuesday, September 21, 2010 4:18 PM > To: Michael Richardson > Cc: Pascal Thubert (pthubert); ROLL WG; IPv6 WG > Subject: Re: [Roll] Flow Label: 12 bits mutable and 8 bits immutable > > On Sep 21, 2010, at 15:48, Michael Richardson wrote: > > > Pascal> We'll be very glad that 6LoWPAN compresses RPL > > Pascal> optimally. But RPL being layer 2 agnostic cannot depend on > > > > Only when the security is provide by the layer 2 can the layer 2 do any > > compression. Otherwise, the encryption of the RPL messages renders the > > compression useless. > > I think Pascal was talking about the encapsulation RPL provides for data packets > (RPL option, draft-ietf-6man-rpl-option-00.txt), not about actual RPL messages. > Do you think the encapsulation information will be encrypted? > > Gruesse, Carsten -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
