Dear RIPE Atlas users,

Some of you may remember earlier discussions about this topic. We’re now ready to enable a feature to address this issue.

Supporting each probe requires resources from the system; at the same time having too many probes in the same network and location does not provide much new information (that is, more probes in the same location or network would provide similar measurement results). Therefore we have controls in place to limit the amount of software probes that are connecting from particular networks:

1. Each probe host can connect up to 2 software probes from the same IP address.
2. Each probe host can connect up to 4 software probes from the same IP prefix (as seen in BGP).
3. At most 32 (or 64) probes from the same IPv4 (or IPv6) prefix (as seen in BGP), regardless of the host.

Software probes that connect from networks beyond these limits will not be allowed to connect, unless and until other software probes drop out. Note that we may also adjust the particular maximums over time.

We realise that in some cases multiple probes belonging to the same user can come seemingly from the same network, violating rule 2 (mostly), yet they arguably still provide enough value for the community. This may be for example because they are in fact geographically widely distributed. In such a case the host can reach out to us via atlas@ripe.net with an explanation and we will consider relaxing this limit.

While in the past we observed surprising corner cases, the particular limits described above are set partially taking into account the current software deployment. Therefore only a handful of probe hosts should be impacted by this (as I write this, 71 probes from 13 hosts). As I mentioned above, we’re open to adjust the limits if needed.

Regards,
Francesco Iannuzzelli