Felix Baumgarten and the ghost in the machine

When Felix peaked…

Every so often it’s nice to have a break from BT network failures, from cyberthreats and handwringing over state aid, and think instead of Gee Whiz! moments like Felix Baumgarten’s 39km, 10 minute plummet to break the sound barrier, now immortalised in Lego, on 14 October.

As Wikipedia will tell you (please support it with a £5 donation), it was also the 65th anniversary of Chuck (The Right Stuff) Yeager’s rocket-powered breaking of the sound barrier.

The fact that Felix flew/fell from Roswell, New Mexico, will of course delight space travel conspiracy theorists.

According to some sources, Felix’s jump was the most watched online event in history. Ripe NCC, the guys who help manage the internet in Europe, and who may be out of a job of the ITU and telco trade association ETNO have their way at the WCIT talks in Dubai this week, measured internet traffic at a number of popular internet peering points (IXPs) during Felix’s rise and fall from the Red Bull Stratos balloon.

Felix’s ascent started at around 15.30 UTC; he leaped into history at 18.07, and landed safely 10 minutes later, having averaged 234kmh, peaking at 1,343kmh.

As you would expect, net traffic at most IXPs showed a noticeable surge during the flight, but the Ripe NCC chaps explain that what they measured was likely a fraction of the actual traffic generated by the event.

This is because of content delivery networks (CDNs). These are essentially big server farms that live at the edge of telcos’ core networks and cache copies of popular content. YouTube and Akamai are examples of CDNs. Because CDNs live at the edge of core networks, anyone connected to the same local network as the CDN gets very fast delivery of their content, be it a Netflix movie, a Microsoft update, or Felix’s Fall. They don’t have to wait for the content to go through an IXP.

Instead of having a single server in Roswell feeding all seven million folk watching Felix at once, it fed a few CDNs around the world, which Ripe NCC picked up. The local CDNs then fed local users, thus minimising traffic across the internet and giving users a better watching experience, especially if you are on rotten ol’ copper.

Even though most users are unaware of them, CDNs sit between the telco and the end users, which is why CDNs don’t really shout about it. There’s a lot of talk now (not unrelated to ETNO’s proposals) about whether telcos should simply carry CDN traffic or pair up with CDN operators, or get into the business themselves.

Given that BT has spent around £1bn to buy the rights to show some sports matches online, what it needs to do should be obvious. But let’s see what actually happens.