Re: [v6ops] PMTUD issue discussion

[Fernando Gont <[e-mail]>]

On 09/04/2014 05:22 PM, Fred Baker (fred) wrote:
>> 
>> I can only think of less than a dozen that I would expect to see,
>> assuming that the actual MTU is based solely on the underlying
>> network technology.
> 
> There are a number of magic numbers in the literature, notably the
> values in RFC 2470, which is for IEEE 802.5 Token Ring. As a
> practical matter today, I think the magic numbers are 1500 and 9000,
> and those numbers reduced by the sizes of various encapsulations. I’m
> not quite sure of the math that led us to select the number 1280, but
> I’ll certainly agree that any encapsulation I can readily think of
> would work with a packet of that size.

FWIW, MTUs of around 296 bytes are reportedly quite usual in some radio
networks. That's why some operating systems enforce that as the minimum
acceptable reported MTU in ICMPv4 "frag needed but DF bit set".

What happens when/if such lins employ v6 (whether they really offer a
"virtual" MTU >=1280).. I just don't know.

Thanks,
-- 
Fernando Gont
e-mail: [e-mail] || [e-mail]
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1