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

["Templin, Fred L" <[e-mail]>]

Iljitsch,

You may be missing the "dumbbell configutation" where
both the IPv6 host and translator are on 9KB links but
there are 1500 links in between?

Fred
[e-mail]

> -----Original Message-----
> From: Iljitsch van Beijnum [mailto:[e-mail]]
> Sent: Tuesday, February 09, 2010 10:18 AM
> To: Dan Wing
> Cc: Templin, Fred L; [e-mail]
> Subject: Re: [BEHAVE] Fwd: IPv6 hosts sending <1280 byte packets
> 
> On 9 feb 2010, at 19:09, Dan Wing wrote:
> 
> > Which is why tweaking the TCP MSS is a helpful workaround
> > to blocked ICMP PTB.
> 
> There very little need to do anything to the MSS option, see my previous message.
> 
> Only in the case where the IPv6 host supports an MTU larger than 1500 but the translator doesn't,
> there is a possible optimization by having the translator rewrite the MSS to the maximum that the
> translator supports so the translator wouldn't have to send too bigs and there is less risk of PMTUD
> black holes. However, it's not clear to me that this case is worth the trouble.
> 
> > But tweaking TCP MSS restricts us to
> > the lowest common denominator (that is, limiting TCP MSS
> > means we can't expect to use larger-than-1500 frames).
> 
> Obviously you would be advertising the minimum of what the IPv6 host and the translator can handle
> and not simply hardcode 1500. If both do more than 1500 bytes then you advertise more than 1500 bytes
> and you get to use more than 1500 bytes. Note that this won't trigger any black holes because the
> IPv4 hosts will send 1500-byte packets. (The intersection between users with broken PMTUD setups and
> users with an MTU larger than 1500 is rather small.)