[atlas] Identifying widespread outages using atlas?

Hi Rafał,

On 20/07/16 16:03, Rafal.Jankowski at thomsonreuters.com wrote:
> Emile,
> 
> Thank you for your response, that's lot of interesting reading.
> 
> By the way I tried to follow one of the links from your articles: 
> https://labs.ripe.net/Members/cristel_pelsser/pinpointing-delay-and-forwarding-anomalies-in-ripe-atlas-built-in-measurements
>
> but it seems to have endless HTTP redirection loop (screenshot attached)

Apologies for that. I'll ask our web team to take a look at this.

> 
> So please correct me if I'm wrong but except for the
> http://atlas-stream.ripe.net API there is no tool for end user to
> detect massive disruptions but you are working on it and can help
> with investigations if you find the case interesting? What would be

Yes, that sounds like a good summary. Others will comment if i missed
something of course.

We had some nice hackathon projects on visualising probe disconnects
(for instance
https://github.com/RIPE-Atlas-Community/ripe-atlas-halo/tree/master/src/halo),
but as far as i understand these took the raw signal and didn't do any
additional statistical processing, which is something we have under
research currently.

> the good address to discuss observed Internet outages, would it be
> this mailing list?

There is a mailing list dedicated to discuss Internet outages
(https://puck.nether.net/mailman/listinfo/outages); the specific GB
outage you referred to earlier has been discussed there some:
https://puck.nether.net/pipermail/outages/2016-July/009281.html

> By the way what is the preferred quoting style here. Is it top
> posting or is it better to reply inline? Is it ok to use RTF/HTML
> formatted e-mails or should we keep them in clean plain text?

I prefer inline, and plain text, and i have a sense that many share that
preference.

cheers,
emile