Reliable neutral global networks for connectivity testing

Mel Beckman mel at beckman.org
Fri May 5 17:00:07 UTC 2023


Most backbone providers operate networks like this internally for their own SLA monitoring, which you can purchase, but what motivation would there be for someone to offer such services for free? After all, you can’t put an ad on a ping :-)

 -mel beckman

On May 5, 2023, at 9:43 AM, Dmitriy A. <dak at prospectone.io> wrote:


Hey all, I have a niche problem and wanted to get some feedback and ideas.

I am trying to find some global networks that accept ICMP requests that I can use to ping and based on the packet loss try to understand if the client is having connectivity issues or uses an unreliable internet connection. The exact testing logic is here https://github.com/jsdelivr/globalping-probe/blob/master/src/lib/status-manager.ts#L66
It basically requires 2/3 endpoints to have 0 packet loss when sending 6 packets to each.

Some requirements:
- Must be a global anycasted network since the client devices can be anywhere
- Must be neutral. Meaning no government or ISP having a reason to block it. This eliminates Google, Microsoft, Akamai, Cloudflare... (blocked in China and elsewhere)
- Must be reliable to avoid false-positives as much as possible.
- Must be public and open. The network must be ok with random devices sending packets to them
- Must be safe and controlled by a trusted entity

I am currently trying to use DNS root servers https://root-servers.org/ but turns out many of them are unreliable for this use-case, I have noticed bad routing resulting in high latency and lots of packet loss at random.

Same goes for pool.ntp.org<http://pool.ntp.org> where some Chinese users hit Amsterdam for some reason. This results in high packet loss marking the test as false-positive.

Popular public DNS resolvers are also controlled by companies that are mostly likely to be banned in certain countries.

Am I missing any other networks that fit the requirements? Any ideas on how to go about this would be appreciated,

Thanks
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nanog.org/pipermail/nanog/attachments/20230505/a146930a/attachment.html>


More information about the NANOG mailing list