[as112-ops] IPv6 network block for AS112 and experiments

[William F. Maton]

On Thu, 14 Jul 2011, Clement Cavadore wrote:

> On Thu, 2011-07-14 at 14:33 -0400, William F. Maton wrote:
>> Clement:  I don't think it's any harm, as it will be interesting to see to
>> which path the IPv6 traffic settles on for its destination. :-)
>
> Great. Regarding the node we run @Hivane/34019, we are in the process of
> connecting it directly to some parisian IXes. At the moment, we have
> contacted FranceIX, Equinix-Paris, and Sfinx, and FranceIX accepted the
> connexion.

Good stuff!

>> It might be a useful exercise to map out where the traffic borderlines are
>> with the two nodes currently reachable via IPv6.  I had forgotten that one
>> of the IPv6 upstreams for my node is CANARIE, which then went off and get
>> free IPv6 transit for itself and downstreams, so I only expected to get as
>> far as the research networks.
>
> How do you think to proceed ? Using various lg/route-servers ?

I've gotten to the point that I'm too lazy to go look these up through 
route-views.  So now I'm writing a program that will go off and fetch 
those, ostensibly to help update the official list.  The data it pulls 
could be used to make a limited view of the public AS112 landscape.

>> The DNS-OARC presnetation from a couple of years ago about locating AS112
>> nodes had a couple of slides that demonstrated odd paths and why having
>> more AS112 nodes closer to their bad-boy sources is a bonus.
>
> It sure may be a bonus. However, routing seems to be weird. For example,
> in Paris, we receive almost much traffic from... .kr ! :-)

That's not a good thing.  I suspect - no, I know that if we get an 
actual IPv6 delegation between our two servers, some of those lookups are 
bound to take IPv6 tunnels (I remember the state of IPv6 connectivity an 
.kr) and the whole thing just goes down-hill after that....

I'm wondering if there isn't some way to satisfy the theoretical with 
doing a practical, limited-time trial and having just one IPv6 .arpa 
subnet delegated at us to confirm this?

wfms