On 12/11/2018 14:04, Robert Kisteleki wrote:
We are indeed looking at what it takes to do RIPE Atlas "software probes". From the hosts' perspective this is likely going to be a software package, or docker image, or such, that one can install on an existing device.
I suggest looking into snap packages, too. As part of your evaluation on the distribution method of 'software probes'. The plus regarding snap packages is they are automatically updated without the need for user actions. (The auto-update functionality being a core design principle with the snap package manager.) Resulting in always up-to-date 'software probes' running on $ENDUSER machines.
From our perspective this is a different approach from what we have so far (hardware-only probes), so deciding on if and how to expand to this area is not a trivial question. VM anchors are simpler in this regard.
As of now we're checking what software probes would change in terms of how we manage our infratructure and reviewing/verifying the related processes. I expect that we'll do some kind of pilot early next year, then decide if this is indeed a desired direction for RIPE Atlas.
I hope that answers your question. Let me know if you need me to elaboration more.
Kind regards, Christoffer