I recently connected a small device to my network that I received from the RIPE Atlas project and in doing so have become part of something pretty cool. RIPE Atlas is open to anyone to join, view, and collect vast quantities of measurements and metrics from all across the Internet.
Statistics are cool
There's something fascinating about statistics, I don't know if it's the sheer scale to which they can extend or how many ways there can be to interpret them. Whatever the case, they can be really useful and interesting. A few months ago, I was talking to someone which lead to the question "How fast is the Internet?" being asked. Now, that's not a particularly answerable question but it is a very interesting thing to think about. If you see the internet as a lot of individual networks connected to each other then here's a virtual high five. But this far from simple structure means that for starters, what on earth do you measure? Where are you measuring from and what are you measuring to?
(How about using any of those RIPE Atlas probes for the from?)
I've recently put a lot of time and effort into improving my systems and practices for every IT system I manage. There's all sorts of things to consider and they all ultimately aim to keep everything running and running well. If something of mine were to break, then it not only is annoying for anyone who notices but it reflects negatively on my reputation. People trust me to manage systems for them and although it's basically impossible to keep everything working perfectly 100% of the time, if handled poorly then the effect any problem, however small, would have on my reputation can be greatly amplified.
You may have noticed the "for anyone who notices" bit in the last paragraph. It's an interesting point, how are you supposed to fix something if you don't know there's a problem? Wait until someone notices AND tells you about it? Well, that could take hours or even days. The answer is monitoring, lots and lots of monitoring. You don't just want to know if something isn't working at all, you want to know of any issue, affecting any number of people - you probably want to know even if it's not affecting anyone. Most monitoring methodologies only probe systems from one or a handful of places. How are you supposed to know that Country X or even ISP X in village X can't access your website?
That picture is a map of 1000~ probes that tried to ping this blog during a recent outage at OVH (server hosting provider), it's a great example usage for Atlas and clearly shows how severe the problems were. You can also get a fair idea of how the problems were affecting them because the UK and most of France were still able to get through. Anyway, this is about the point when I reveal more details about the magical solution to all life’s woes... Well, maybe not all of life's woes (maybe some other time...) but at least for the problems I've spoken about here.
What is RIPE Atlas and who is behind it?
RIPE NNC is the Réseaux IP Européens (French for European IP Networks) Network Coordination Centre for Europe, the Middle East and parts of Central Asia. It is one of 5 Regional Internet Registries (RIRs) who each manage Internet resource allocations (IP addresses for instance) as well as registration services and various other things that support operation of the Internet.
One of RIPE's projects is called Atlas and is a global network of probes that measure Internet connectivity and reachability. Atlas is a global network of probes hosted by volunteers (such as myself). They have just passed 10,000 of these probes active globally - a very impressive feat.
(A picture of the probe I received, RIPE's v3 probe which is based on a TP Link wireless router)
I received the probe for free via their website where you can request a probe to host. The Atlas project is made up of users (someone who accesses the RIPE Atlas maps and stats), hosts (someone who connects a probe or anchor to their own network), a sponsor (individual or organisation who support RIPE Atlas financially) and finally, ambassadors (which is what I have just become, someone who helps distribute probes).
Naturally, I couldn't help but have a look at what the probes "use" in terms of traffic and power consumption, my probe uses the following:
- Peak of 10 Kbps (around 7 Kbps average) for a total of up to 3 GB per (31 day) month
- .16 amps at 5v (it's USB powered) for a total of around 600 Wh per (31 day) month which equates to 6p at 10p per kWh, not bad eh?
So far, we know there's a probe you can get from an organisation called RIPE and host to support the project. This may be enough for some of us to dive in but RIPE don't expected you to help for nothing - hosting a probe earns you credits which can be used to run your own measurements using ANY of the probes in the network (you can use up to 1000 probes per measurement). I love this because it's useful for someone like me from a production environment perspective, but it also provides a great way to learn about the topology and functionality of the Internet (and networks in general). It also shows you a ton of information and words that you may well not know about or understand, giving you a great opportunity to learn about them and understand their place in the real world. Of course, you don't have to do anything with a probe other than plug it in and create an account. If you feel generous, you could send your credits to someone else.
Why I chose to host a probe and why you should consider doing so too
I could simply answer this with "I was curious" but that wouldn't be awfully interesting. I'm a huge fan of community driven and open projects. RIPE have done something really impressive with Atlas, they've created hardware probes that can be connected and forgotten about and a great online interface for viewing and creating measurements which can be accessed by anyone. Simply being able to say, "I'm a part of that" feels pretty cool.
You can use RIPE Atlas to measure Internet outages, to run measurements using DNS, Ping, Traceroute, HTTP, SSL and NTP against a host or hosts of your choosing, or to simply allow others to run measurements using your connection, serving as yet another vantage point on the Internet for you and others to use. I've got a few ideas for interesting uses for Atlas which I may well write about in the future.
(A colour coded map showing the latency from 1000 probes to this blog, no prizes for guessing which country it's hosted in; if you can picture what is behind the dots, that is.)
So if you want to be part of the world's largest Internet measurement network then pop over to the RIPE Atlas site. As I'm now an Atlas ambassador, there's a delivery of a few probes on its way to me, I'll see how many they send me and update this post if I have any spare.
- RIPE Atlas Brochure: /download/RIPE-Atlas-2015.pdf
- RIPE Atlas: https://atlas.ripe.net/
- RIPE Coverage Statistics and Map https://atlas.ripe.net/results/maps/network-coverage/
Short link: on-te.ch/atlas