[ipv6-hackathon] Combining TraceMON & project proposal 5a at ipv6-hackathon?

Hi,


On 03/11/2017 08:54, Vesna Manojlovic wrote:
> Hi,
>
> while reviewing projects with Pedro, we thought that this might be
> possible to combine with some of the already existing features in TraceMON:
>
>> 5a. Comparing IPv4 and IPv6 traceroute paths
>> by Petros Gigis
>>
>> Create a visualization to compare IPv4 and IPv6 traceroute paths, highlighting the differences between these paths. For example, it would be interesting to observe differences in IP-level/AS-level paths in terms of traffic locality; e.g., IPv4 paths from ASX to ASY staying inside a country while IPv6 paths are leaving it. Moreover, this visualization would help the network operators and the academic community understand the differences in IPv4 vs. IPv6 routing, on the data plane. Furthermore, some kind of statistics could be created based on the visualization (similar to the Eyeball Jedi approach (sg-pub.ripe.net/petros/eyeball-jedi) to compare country ecosystems at a larger scale.

Commenting on this part:
>> Finally, comparing the IPv4 ASN of the endpoint probes (as given by the RIPE Atlas platform) with their IPv6 ASNs, would potentially reveal native IPv6 support on the AS vs. support from IPv6 providers.

My approach for comparison is usually to ignore probes where their v4 
and v6 ASNs differ.

Most commonly where they differ, the v6 side is Hurricane Electric so 
realistically it's not going to offer a useful comparison. There's a 
long tail of other v6 ASNs after HE, but IIRC it's a relatively small 
set. (I bet many of those ASNs actually belong to the same organisation; 
having these smarts in tracemon or similar would be very useful.)

One thing that traceroutes will have a weird time with is transition 
tech: nat64 if the probes work behind that, or ds-lite, most commonly. 
It would be neat to identify or hint at those.

> Massimo & Emile, could you care to comment?
>
> Petros, Steve, Jasper & Pedro,
> and aevryone else on site tomorrow, interested in this project - let's
> consider TraceMON...
>
> https://labs.ripe.net/Members/massimo_candela/tracemon-traceroute-visualisation-network-debugging-tool
>
> https://github.com/RIPE-NCC/tracemon

Yes, it'd be great to show traceroutes from the same probe to the same 
target side-by-side. (But I'm not the person you want writing this code.)

For visualisation, what is really vital is aggregates. Let's say I'm an 
ISP or a CDN, what I often care about is aggregate traffic. A 
query-based system ought to allow an operator to select *all* 
traceroutes from $MY_ASN to $ANOTHER_ASN over a time period, and 
aggregate the paths together, and able to flip between v4/v6/both.

An amazing direction to go would be towards transit maps where line 
width indicates frequency or volume, similar to:
- https://www.cambooth.net/project-france-tgv-routes/
- https://twitter.com/lisacrost/status/917852662980382720

That sounds like a much bigger project than a hackathon weekend, but I 
think it's where the value is.

S.


>
> Thanks,
> Vesna
>
> _______________________________________________
> ipv6-hackathon mailing list
> ipv6-hackathon at ripe.net
> https://lists.ripe.net/mailman/listinfo/ipv6-hackathon