[afnog] Survey: Collecting African IXP Colocation Data for research purposes (Slight changes in the survey)

[Roderick]

Hello Nishal, SM and all, 

First the RTTs among source and destination IPs is really high (over 200ms) hinting an intercontinental path.

As per the IP to AS mapping and IP geolocation (with maxmind) the corresponding AS  and CC paths are the following:

IPs -> AS ASName -> CC
197.226.251.86 &196.20.254.173 -> AS23889 Mauritius Telecom, MU ->  MU
80.231.76.61 & 80.231.76.86  -> AS6453 TATA, US -> EU
130.117.15.177 & 130.117.49.86 & 149.6.148.202 -> AS174 Cogent, US -> EU
41.76.41.174 & 41.76.41.174 -> AS30999, EMTEL-AS-AP, MU -> MU
196.192.112.67 & 196.112.83 ->  AS37708, AFRINIC, MU -> MU

Mauritius Telecoms is clearly transiting via TATA and Cogent in Europe to reach EMTEL, although both are in the same country. AFRINIC which appear to be the AS destination is a client of EMTEL (https://stat.ripe.net/AS37708#tabId=routing).
There is clear evidence that this path is not passing via a local IXP. And this is confirmed by the country path.

I advice those networks to do so. Because, if AS23889 and AS30999 were peering at the Mauritius IX, then the Mauritius IX peering LAN (that I still do not know) would have 
appeared in the IP path instead of ASes 6453 and 174,  the country path would have been [MU, MU, MU …], the RTT would be really low, users would have experienced 
better QoE and both ISPs would have saved on transit costs. 

As per your previous mail, I define an IXP as a layer 2 or 3 switch where 3 or more ISPs (local or not) interconnect their network to exchange traffic or access to common services hosted there.
So according to me, the concept of having an IXP per country where only local networks interconnect is a case proper to Africa, due to the fact that the Internet is not opened in the region: this should change.

- If ASes 23889 and 30999 are listed to me as “the only peer at the IXP”, I would not consider Mauritian IXP as an IXP (without even performing traceroutes and carrying an AS path analysis). 

- If  “3 peers" among which both ASes 23889 and 30999 are listed, then given the AS path above the Mauritian IXP is not working; since all networks should be peering with one another. 

- If more than 3 peers are listed to me, then this only traceroute is not sufficient to decide; since an ISP can decide to peer with some members and not with others. 
  Please note that traceroutes should be performed in both directions and note that there are 30 ASNs in Mauritius (https://stat.ripe.net/MU#tabId=database). 

We have tried this exercise by helping deploying probes within local networks, performing full mesh traceroutes, and checking if local networks are peering. Unfortunately, probes are not in all networks and can never be. 
The only thing that can help us decide is an input from existing IXPs.  

And with this example, I now think our methodology would be more trustful if we suppress from the list, all IXes that do not reply to the survey. Thank you for the comment. 

I hope this helps, 

BR, 

Roderick






On Feb 28, 2016, at 1:48 PM, Nishal Goburdhan <nishal at controlfreak.co.za> wrote:

> On 28 Feb 2016, at 23:12, SM wrote:
> 
>> Hi Roderick,
>> At 10:32 26-02-2016, Roderick wrote:
>>> MU; Mauritius; MUIXP;
>> 
>> 197.226.251.86  1.262 ms  1.419 ms  1.635 ms
>> 196.20.254.173  4.268 ms  5.87 ms  4.389 ms
>> 80.231.76.61  209.599 ms  213.320 ms  297.229 ms
>> 80.231.76.86  210.282 ms  212.957 ms  210.336 ms
>> 130.117.15.177  209.689 ms  209.623 ms  209.983 ms
>> 130.117.49.86  210.858 ms  211.10 ms  211.408 ms
>> 149.6.148.202  209.242 ms  209.49 ms  209.279 ms
>> 41.76.47.65  209.756 ms  209.122 ms  209.560 ms
>> 41.76.41.174  211.381 ms  210.837 ms  210.433 ms
>> 196.192.112.67  214.421 ms  214.225 ms  214.91 ms
>> 196.192.112.83  213.649 ms  215.91 ms *
>> 
>> Does that mean that the IXP is working?
> 
> 
> www.mixp.org.
> be your own judge.  
> 
> and vote with your wallet.
> 
> —n.
> 
> _______________________________________________
> afnog mailing list
> https://www.afnog.org/mailman/listinfo/afnog

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.afnog.org/pipermail/afnog/attachments/20160228/1dadb5a8/attachment.html>