Re: [Int-area] I-D Action: draft-ietf-intarea-frag-fragile-00.txt

[Brian E Carpenter <[e-mail]>]

Hi,

Earlier I said:

>>    Application developers SHOULD NOT develop applications that rely on
>>    IPv6 fragmentation
> 
> It isn't obvious to me that this is an algorithmic requirement. If the application
> runs over TCP, how does the developer ensure that TCP will use an MSS that
> avoids the need for fragmentation? (The UDP situation is a bit clearer, but
> RFC8085 introduces considerable complexity, unless we accept that the
> Internet MTU for UDP is 576 or 1280.)

I've looked into this a bit for the UDP case, which doesn't have
the added complexity of MSS negotiation.

The only surefire way of avoiding fragmentation is to switch it
off. The recipe for doing that varies by operating system and version,
and between IPv4 and IPv6. (I have Python code that I've exercised
on Windows 7, Windows 10 and Linux. Thanks to Tom Herbert, Praveen
Balasubramanian and Nick Grifka for advice.)

Whether that automatically solves the problem depends on whether
PMTUD works, which is unpredictable. Therefore, if you want a protocol
*design* that is universal, it can't depend on PMTUD. Disabling 
fragmentation on its own is not enough. (That isn't news.)

The only universal solution therefore seems to be a fully specified
PLPMTUD solution, either built into the protocol design, or available
as a normative reference. I think that would be a better
recommendation than simply saying "don't rely on fragmentation."

IMHO RFC4821 isn't a sufficient normative reference for this.
Possibly draft-ietf-tsvwg-datagram-plpmtud will do it for UDP-based
protocols. But IMHO we need a solution for each transport protocol,
with widely available libraries.

   Brian