Hi,

On 2010/01/19, at 10:26, Brian E Carpenter wrote:

> Rémi,
> 
> On 2010-01-19 01:46, Rémi Denis-Courmont wrote:
>> On Fri, 15 Jan 2010 20:13:16 +0900, Arifumi Matsumoto <[email protected]>
>> wrote:
>>> * Fred's proposal.
>>> - A host tries each pair of src and dst addresses to establish
>>> a connection in a short time period.
>>> - A host can make use of ICMP error messages indicating that the
>>> src address should be this, or the src address is simply wrong.
>>> - A host can have cache for the reachability status, which stores
>>> which src and dst pair should be used to reach a certain dst.
>> 
>> This is awfully complicated. It might be implemented properly in some
>> high-profile apps, but it will be implemented incorrectly in many apps, and
>> it won't be implemented at all in others. 
> 
> It really doesn't belong in apps at all. It belongs in a shim common
> to all transports. I would suggest giving serious consideration
> to modifying the spec for the REAP component of SHIM6, which incidentally
> doesn't rely on ICMP transparency.

We can implement this try-and-error approach mechanism in several ways,
and each of them has its own benefits and limitations.

And as Brian suggested, we can learn from the past/ongoing activity
like Windows APIs, SHIM6 and mptcp.


> However, it seems evident to me that we have to both make
> site-wide 3484 policy distribution work, and make rapid trial-and-error
> work. Neither one alone meets the need.

Same here.

Kindest regards,

> To this day, there are still
>> applications that won't even try to fall-back from IPv6 to IPv4. Also, as
>> you mention, this simply doesn't fly for connection-less protocols. So
>> IMHO, source address selection should be hidden from apps into the
>> operating system as much as possible.
>> 
>> That of course does not preclude writing advices for robust applications on
>> what intelligent things they can and can *not* do. That could definitely
>> include trying all potential pairs in parallel instead of trying them
>> sequentially, if the application-layer protocol would handle it properly
>> anyway.
>> 
> 


--
Arifumi Matsumoto
  Secure Communication Project
  NTT Information Sharing Platform Laboratories
  E-mail: [email protected]

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

Reply via email to