RIPE71 (2)

Two days and a big pile of notes later, it's time to write about what happened Tuesday and Wednesday here at RIPE71. There's no replacement for actually being here (and getting to talk to so many smart people) but if you couldn't make it then all talks are recorded and made available online, together with the slides and transcripts.

I've also written about the first day here and you can also read a different viewpoint on the ISOC's Deploy360 blog.


Early start for the only full day of plenary sessions with Dario Rossi's Where are the Anycasters?, talking about geolocating anycast IP addresses. One of my favorite talks of the day, unfortunately missed by all the late sleepers that didn't make it at 9am. He talks about how they detected anycast IPs (one way is to look for latency anomalies such as 2ms from Europe to an US IP) and then how they actually manage to pinpoint their location on the map. I won't write any more as I couldn't do it justice in a few words.

It was measurement morning, so we heard about the MONROE project (Measuring Mobile Broadband Networks in Europe) that aims to provide a hardware platform (with Linux) for running experiments on mobile networks and generating reports and statistics, usable by both operators and end-users. Time to move from Europe to New Zealand, with Jamie Horrell talking about his mapping of NZ's broadband infrastructure. A lot of data, a lot of politics and really good results.

A break, some snacks and coffee and it's time for automation! And datacenters! And kittens and puppies! To start off, Jose Leitao and David Rothera talked about Dr NMS or: How Facebook Learned to Stop Worrying and Love the Network - it's the third time I see this (UKNOF and iNOG::3), these guys are touring hard. There's some good stuff in there, but for me it's a bit too high level - I'd like more details about how they solved some of those problems, but you can't have it all.

Next up, the best talk of the day, aptly named and scheduled: Building a Small DC... For the rest of us by Karl Brumund. I quote, "Cause we're not all Facebook, Google, Amazon, Microsoft...". Yeah, pure genius. But some sense of humour a great talk does not make - it's the content and the lessons learned that are pure bliss for our real life battered engineer souls.

And the final piece of the automation puzzle, after we heard about how others do it at large scale (FB) and at a small scale (Dyn) - it's time to learn how to start on that path with Leslie Carr's What is NetDevOps? Why? talk. Well played RIPE PC, well played.

Leslie talks and shows examples of the basics of using things such as git, puppet/chef/ansible, travis-ci/jenkins and how to get started on that road to doing less of the monkey work manually and trying to avoid fat-finger mistakes.

Also worth mentioning is one rather special non-technical topic, where a panel of law students from Leiden University came to talk about The Internet as a Field of Modern War: The Applicability of International Humanitarian Law to Cyber Warfare. I got to chat with Louis (one of the presenters) at the welcome drinks and I think they're in uncharted territory, but a very interesting one at that.

If you're interested in mobile trends and IPv6, then watch George Michaelson's talk A Look Under the Hood at Devices, Networks and IPv6 and if you still have time the lightning talk on Measuring Multi-CDNs is worth watching.


As we're almost done with plenaries, it's a full day of Work Group sessions, split between both rooms. Lots of interesting things to discuss with some topics becoming quite specialized.

Address Policy WG

After the crazy ride that Tuesday was, I'm happy that today I got to sleep a bit later, but I made it just in time to catch Ciprian Nica explain Romania's Jump to 1st Place as Exporter of IPs. It's a staggering amount that we've sold (about 5 million IPs out of a total of 13.5 million), but they were mostly IPs that had been rented out to companies that turned out to run a spam-sender-IP rental business. It's quite interesting that in a 2012 report they found out only about 4.2 million IPs were actually used by the major ISPs and mobile providers for their customers (accounting for basically ~ 90% of the market).

Connect WG

The MANRS (read manners) presentation generated quite a bit of debate at the end, with one clear outcome: some operators don't have time/budget/interest in creating tools to deal with the problems of the Internet community, so something should be ready-made that can be given to them as a "just take it and implement it, it'll do you (and everybody else) a lot of good". Tough one.

Just before the meeting a RIPE Atlas hackathon was organized and ASNtryst is one little project that came out of it - it's short and very cool showcase of what a few people can quickly put together by simply analyzing readily available data.


IPv6 and the Enterprise was a story of the sad state of things in the enterprise, with a very realistic view of how the TCP/IP model actually looks like and how all the other additional human layers come on top.

DHCPKit is a "when it doesn't work properly roll your own and open source it" project that implements a DHCP server in python with a philosophy that reminds me of ExaBGP. Apparently I'm not the only one.

Flow-based Detection of IPv6-specific Threats is a project that aims to create flow-level signatures of IPv6 specific threats, so that existing stat crunching infrastructure can look at some of the inevitable nasty stuff that's going to come via IPv6 (in case you thought the baddies only stick to IPv4). The problem is vendor support of the various new fields in the flow specs, but it's something that needs to be addressed or provided by the community as plug-ins.

See you tomorrow...

One full day of WG sessions coming up (Routing, DNS, IPv6, Open Source) and the RIPE Dinner at the Palace of the Parliament. If you're around and I haven't said hello yet, hopefully I'll see you there!

And, as always, thanks for reading.

comments powered by Disqus