j...@mercury.lcs.mit.edu (Noel Chiappa) writes: > The UDP checksum in the outer header on LISP user-data does nothing, is > expensive/impossible to compute (depending on the hardware), and therefore the > correct practical engineering choice is to not compute it.
You will probably eventually see PC-based routers which verify all checksums and drop invalid ones before letting the CPU see them, if the IPv6 specs allow it. PC NIC's are optimized for host use, not for routing, and therefore they look at the whole packets anyway. Most of them probably don't do much for IPv6 yet, but that is only a matter of time. Now such routers may forever be sufficiently rare on the open Internet that it is not a concern for LISP deployment. It seems prudent to try to get a formal exception from the usual UDP checksum rules if you really want to go for the no-checksum option. Personally I would drop zero checksum UDP packets with a rate-limited ICMP error when doing the LISP encapsulation. I would expect the number of zero-checksum UDP packets to drop as more NIC's get the checksum feature. /Benny -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
