Hi Brian, 

> -----Original Message-----
> From: Brian E Carpenter [mailto:[email protected]] 
> Sent: Wednesday, January 04, 2012 6:55 PM
> To: Templin, Fred L
> Cc: Fernando Gont; [email protected]
> Subject: Re: Fragmentation-related security issues
> 
> On 2012-01-05 11:50, Templin, Fred L wrote:
> >  
> > 
> >> -----Original Message-----
> >> From: Fernando Gont [mailto:[email protected]] 
> >> Sent: Wednesday, January 04, 2012 2:22 PM
> >> To: Templin, Fred L
> >> Cc: Brian E Carpenter; [email protected]
> >> Subject: Re: Fragmentation-related security issues
> >>
> >> On 01/04/2012 07:06 PM, Templin, Fred L wrote:
> >>>> I see no reason to expect that PMTUD will be more reliable for
> >>>> IPv6 than for IPv4.
> >>> I think a lot is now hinging on the assumption that
> >>> PMTUD for IPv6 works. Unlike the situation for IPv4,
> >>> I see no reason to expect that PMTUD for IPv6 will
> >>> be unreliable.
> >> It has been found to break, already -- as a result of firewalls
> >> filtering ICMPv6 messages, are some intermediate systems with
> >> inappropriate rate limiting for all ICMPv6 traffic.
> > 
> > If IPv6 PMTUD breaks, it is due to violations of the specs.
> > If IPv6 PMTUD breaks, then we are lost - time to give up
> > and design a different protocol?
> 
> The point is that paranoid firewalls will turn this into an
> arms race - if they are paranoid enough to block ICMP PTB,
> which apparently many are, why wouldn't they block any other
> signalling mechanism - especially a new one?

SEAL provides a new signalling mechanism called "SCMP"
which is intended to traverse firewalls that might block
ICMP messages. SCMP messages include a message signature
that the source node can use to determine whether the
packet-in-error corresponds to a packet the node actually
sent. Under what reasonable circumstances could even a
paranoid firewall block that?

> That's why RFC 4821 describes MTU probing hidden in the transport
> layer, where hopefully firewalls would let it be. You will
> probably look in vain for widely deployed versions of RFC 4821.

Thanks for mentioning RFC4821 - it occurred to me last
nite that I had mis-spoken and that RFC4821 is advisable
even if the network is making a best effort to return
ICMPs. I think this is fairly well captured in Section
5.6 of RFC6434. However, the use or non-use of RFC4821
does not excuse the network from making a best effort
to return the required *CMPs.

Fred
[email protected]

>      Brian
> 
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to