Authors,

I have already worked out the details of the algorithms your draft
is seeking in SEAL:

http://datatracker.ietf.org/doc/draft-templin-intarea-seal/

There are exactly two differences between SEAL and GRE:

  - SEAL does not use IP frag/reass; it uses its own frag/reass
    mechanisms at a layer above IP

  - SEAL requires that the tunnel egress reassemble enough so
    that all packets that are no larger than 1500 bytes will
    be accommodated unconditionally. With GRE, the best the
    tunnel ingress can assume is that the egress has a 1500
    byte reassembly buffer, which is not big enough to hold
    a 1500 byte packet plus the encapsulation overhead

I will say again that for IPv6 the GRE tunnel MUST accommodate
packet sizes up to 1280 as specified in RFC2460. And, when the
path over which the tunnel is configured only has a 1280 MTU
the GRE tunnel MUST use IP frag/reass to meet the IPv6 minMTU
requirement. But, we are seeing in other discussions that both
IPv6 frag/reass and IPv6 PMTUD may not work very well in some
environments.

Thanks - Fred
[email protected]


_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to