If we give a recommendation, I think it should be based on
measured data. See for example Sec. 5.5 of
http://nlnetlabs.nl/downloads/publications/pmtu-black-holes-msc-thesis.pdf
Thanks for the link. And only 1 (out of >1150) probes found a PMTU in
IPv4 smaller than 1280, and that was 1280. More than 2/3 were exactly
1500, and the vast majority was over 1400.
Smallest they found was 1240. Any reason to set the limit (much)
below that?
s/and that was 1280/and that was 1240/
If we think this experiment is serious (e.g. that it covers much of the
Internet, note that 1150 probes is a very small number), then we can
recommend 1280 and cite this reference. Alternatively, we could ask the
folks at TSVWG or IPPM.
I am not familiar with current large scale Internet measurement efforts.
But you an I have one such project nearby: http://www.netdimes.org/new/.
I will ping them.
Thanks,
Yaron
_______________________________________________
IPsec mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ipsec
