Carsten Bormann a écrit :
Alex,
there are lots of IP-over-foo documents that have much more logic than
6LoWPAN.
-- the PPP series of documents,
Right... PPP, EAP and PANA.
I think they can be used below IP _and_ above IP (EAP over 802.1x, EAP
over Diameter; PPP over rs232, PPP over SSH; etc.) PANA always over UDP.
In this vein, could the 6LoWPAN adaptation layer too run somewhere above
IP? If yes - wouldn't it be more efficient for the IP layer to perform
the necessary fragmentation instead?
which by reference includes functions
like ROHC (which has its own segmentation layer);
I think ROHC also runs over application layers too. I think it _mainly_
runs above IP.
-- IP over ATM?
I think IP over ATM has not much logic in it. I see constants
everywhere, no new logic. (by "logic" we both mean some behaviour to be
implemented, more than just mappings of constants and definitions).
-- and how about MPLS (which also includes ROHC via RFC 4901);
I can't comment on MPLS... I have yet to see it running on IPv6.
-- and an example for something quite similar to 6LoWPAN (including
fragmentation), RFC 3146 (IPv6 over Firewire)/RFC 2734 (IP over Firewire).
IPv6 over Firewire saying it performs the same fragmentation as IPv4
over Firewire risks being a long stretch, because IPv6 and IPv4 behave
differently with respect to fragmentation at IP layer.
This is why I don't understand the goal of specifying 6LoWPAN
fragmentation at a layer situated below the IPv6 layer.
It is a bit late to make that argument:
RFC 4919 (defining that goal) and RFC 4944 (specifying the protocol)
were approved in 2007.
Right... these being RFCs requesting comments: I offered my comments.
Alex
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
