Re: I-D ACTION:draft-jabley-ipv6-rh0-is-evil-00.txt

["Ebalard, Arnaud" <[e-mail]>]

Le 11 mai 07 à 23:18, David Malone a écrit :

> On Fri, May 11, 2007 at 02:16:41PM +0200, Guillaume Valadon /
> ???????????? ???????????? wrote:
>> Except some custom-made traceroute6 and KAME's implementation, I am
>> not aware of such usage of RH0. What I mean here, is that deprecating
>
>> RH0 won't harm anyone (except some reasearchers).
>> Discovering the 'return path' is a really cool feature, but the
>> discussions during the last weeks proved that RH0 is not the right
>> solution. Its benefit is too small comparing to the problem related
>> to RH0.
>
> IMHO, the existance of so many traceroute looking glasses is
> significant evidence that operators (not just researchers) need
> intermediate point traceroute.

When I use a looking glass, this is basically because my proxies  
prevent me to go out directly using ICMP and/or UDP and I really  
don't care about the position of the looking glass as far as it is  
available through my proxies.

Probably a silly search, but anyway : a Google fight between "IPv4  
looking glass" and "IPv6 looking glass" provides respectively 156 000  
and 345 000 answers.

As RH0 was fully available on the IPv6 Internet 2 weeks ago, my  
conclusion (don't hesitate to challenge me on that, guys) is simply  
that there is perhaps just no correlation between RH0 and looking  
glass, i.e. they are not used for the same purposes / by the same  
people.

Also, from my perspective, if I want a remote traceroute via a router  
i control, i just log onto the router and launch it from there. And  
as a maintainer of a (small) network you don't control, I'm just  
happy you are not able to use RH0 in it.

Now, a similar question as already been raised before on the list,  
but I'd like to ask it again more precisely : Who on that list has  
already used RH0 for debugging purposes, and how many times ? I don't  
ask about the people that "like" the mechanism or "find it fun".

> I understand Itojun's suggestion that we can have a RH7 that will
> allow useful source routing without the danger associated with RH0.
> This sounds like a very good idea.  However, realistically, I suspect
> that even if the RH7 standard was fully specified tomorrow, we would
> be waiting more than two years to have it implemented in production
> versions of software that operators are likely to be using.

yes. Likely.

Cheers,

a+

--------------------------------------------------------------------
IETF IPv6 working group mailing list
[e-mail]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------