Jared, I’d love to know more about this. Daniel — let me be clear that I have no intention of endangering RIPE atlas, and your concerns are the same as mine when it comes to the system, as well as all of the probe hosts. That’s why we specifically haven’t used RIPE to power any projects. Hence, I’m extremely interested in whether there’s any way that limited HTTP measurements could exist without being a threat. Sisi
On Jan 13, 2015, at 5:41 PM, Jared Mauch <jared@puck.nether.net> wrote:
On Jan 13, 2015, at 5:28 PM, Daniel Karrenberg <daniel.karrenberg@ripe.net> wrote:
sisi,
<candid>
ripe atlas was built for monitoring the infrastructure. the direction you are suggesting bears a significant risk of destroying ripe atlas because we will be regarded as breaking our agreement with probe hosts and probe hosts will disconnect their probes. i am opposed to risking ripe atlas for the sake of the http measurements; specifically meaurements that are bound to make ripe atlas a target. i am all for doing the work you suggest, but i have to suggest that you find people willing to cooperate while knowing the consequences. if we proceed like you suggest we may very well end up with no useful platform for any purpose.
Perhaps this is a good use-case of having a ‘hosted VM’ type probe? The NLNOG RING is an example of something where it’s a pure full-mesh community but without credits.
If there was something like a software probe, i could ask our systems team to deploy one VM per site we have for these purposes, and it could provide this capability with reasonable constraints.
- Jared