-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Iljitsch van Beijnum wrote:
> On 19 mei 2009, at 23:14, Joe Touch wrote:
> 
>>> Didn't you just suggest to just send enough of the original packet to
>>> make ROHC do its thing? That piece of the packet should be small enough
>>> to always fit.
> 
>> Yes, except that I also suggested sending the packet without any ROHC on
>> it too.
> 
> Yes, but then you would still almost never have full-size packets so the
> ROHC only saves bandwidth, not packets.

I wouldn't say only. The number of packets you save is still in the noise...

> Basically this will buy you a 1
> byte larger MTU, not worth the trouble because the IPsec header already
> exposes you to PMTUD blackholes anyway.

Using ROHC allows you 30 bytes more in the MTU if you try hard, but
causes this sort of hiccup on resync packets as a result. IMO, that's
more trouble than it's worth.

> Ideally, the decapsulator can tell the encapsulator what to do because
> the decapsulator knows about whether it's behind a NAT with
> fragmentation issues. But then there would have to be signaling that
> carries this information.

Yeah - that's what Fred was getting at with SEAL, which has that sort of
signalling.

Joe
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkoTJnkACgkQE5f5cImnZrs4CQCffKV0CLgyitOBk7Eznplq6BYQ
5B0An3YCk6KEqkWMFWdgP+ziPso/wuMJ
=2UDf
-----END PGP SIGNATURE-----

Reply via email to