On 01 Feb 2013, at 09:37 , Brian E Carpenter wrote: > I looked at the ILNP RFCs while we were preparing > draft-carpenter-6man-ug-00, and it seemed to me that > ILNP doesn't in fact use IIDs, but rather a redefinition > of the bottom 64 bits as a Node Identifier [RFC6741].
ILNPv6 changes the *semantics* of the lower 64-bits *from the view of the transport-layer and above*, but is quite careful to keep the syntax unchanged. Keeping the syntax unchanged is crucial to ensuring that ordinary IPv6 routers can/will forward ILNPv6 packets -- without even needing to know that they are ILNPv6 packets. In turn, the ability of ILNPv6 to incrementally-deploy within an already extant IPv6 deployment is crucial. Flag-day transitions are broadly unacceptable for very good operational reasons, so ILNPv6 must remain backwards-compatible with existing IPv6. A great deal of time was spent engineering both ILNPv4 and ILNPv6 to ensure that each would be backwards- compatible with IPv4 and IPv6 infrastructure (respectively). > Doesn't that separate the use of the u/g bits > for ILNP completely from their use in plain IPv6? No, as explained above. > Please have a look at the draft and suggest corrections > if it seems wrong to you. I haven't done yet, due to lack of time, but will put reviewing that onto my to-do list. Thanks, Ran -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
