Re: [BEHAVE] Fwd: IPv6 hosts sending <1280 byte packets

[Iljitsch van Beijnum <[e-mail]>]

On 9 feb 2010, at 1:01, Templin, Fred L wrote:

> The concern with setting DF=0 on the IPv4 side for packets
> that are 1280 or smaller on the IPv6 side is the possibility
> for RFC4963 data corruption if we let it run at line rate.

But how can this be avoided by the translator?

In theory if the IPv6 host sees a < 1280 too big it could deduce that its packets are going to be translated to IPv4 with DF=0 and then slow down to avoid wrapping the ID during a packet lifetime, but I don't see that happening...

If an IPv4 network is stupid enough to use such a small MTU then incorrect reassemly of fragments is a risk they are apparently willing to run.

But again, note that few fragments will happen because the TCP MSS will usually make the IPv6 host send packets that don't require fragmentation. The only two applications that I know of that send high rates of non-TCP packets are RTSP video streaming, which I assume won't be done much from an IPv6 client towards and IPv4 server, and tunneling, which I assume will also be rare across a NAT64. Maybe in the future there will also be UDP BitTorrent, but BitTorrent usually doesn't send too much traffic to any given destination and it uses application layer checksums.