Test your connectivity for World IPv6 Day

[Rémi Després]

Le 7 juin 2011 à 14:07, Marc Blanchet a écrit :

> Le 11-06-07 07:55, Rémi Després a écrit :
>> 
>> Le 7 juin 2011 à 13:26, Marc Blanchet a écrit :
>> 
>>> Le 11-06-07 07:03, Mohacsi Janos a écrit :
>>>> Dear All,
>>>> A possible option would be for develop a new opportunistic PMTUD standard:
>>>> - start from 1280 and increase it until it is possible....
>>> 
>>> see RFC4821.
>> 
>> Agreed.
>> 
>> However, RFC 4821 isn't completely clear as to which PMTU to try first.
> 
> I've heard that it is already implemented on linux. don't know the details yet.
> 
>> As detailed in my last answer to Fred, 1280 seems the best choice, at least in IPv6.
>> Can we agree on this?
> 
> nooo!

Too bad for IPv6-connectivity reliability!
Where ICMPv6 don't come back (it is wrong but happens), some connections may be established and later be unable to send full packets.

> IPv4 typical MTU is 1500.

Yes.
A direct consequence is that IPv6 paths that include IPv6/IPv4 tunnels (e.g. with Tunnel brokers or 6rd) may have MTU's < 1500. 
 
> you are suggesting downgrading IPv6 performance (compared to IPv4) by doing MTU 1280? does not make sense to me.

Many users don't care about a minor optimization, but do care about safe connectivity.  


> PMTUd issues are the exception, not the problem. But the level of exceptions are higher now than it shall be later, since we are really deploying IPv6 now, removing old stuff, fixing as we go.


> Think also about backbone nodes/high-end servers that are able to do 9K MTU. Change to 1280 MTU. no.

I didn't forget them, but please note that I don't propose to abandon PMTU discovery, the tool that permits to discover jumbo MTU's.


Let me quote again excerpts of RFC 4821:
" Since protocols that do not implement PLPMTUD are still subject to
  problems due to ICMP black holes, it may be desirable to limit to
  these protocols to "safe" MTUs likely to work on any path (e.g., 1280
  bytes)."
and:
" As an optimization, it may be appropriate to probe at certain common
  or expected MTU sizes, for example, 1500 bytes for standard Ethernet,
  or 1500 bytes minus header sizes for tunnel protocols."
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

(1500 - 40) may be safer than 1500, but 1280 is THE value that is guaranteed to be safe.

RD


> 
> Marc.
> 
> -- 
> =========
> IETF81 Quebec city: http://ietf81.ca
> IPv6 book: Migrating to IPv6, Wiley. http://www.ipv6book.ca
> Stun/Turn server for VoIP NAT-FW traversal: http://numb.viagenie.ca
> DTN Implementation: http://postellation.viagenie.ca
> NAT64-DNS64 Opensource: http://ecdysis.viagenie.ca
> Space Assigned Number Authority: http://sanaregistry.org