A Wi-Fi/VoIP Phone Booth In the Burning Man Desert

Brad Templeton writes "I, (of EFF/ClariNet/rec.humor.funny) along with Brent Chapman (Majordomo/Building Internet Firewalls) and the satellite dish of John Gilmore (EFF/Cygnus/Cypherpunks/etc.) put together an engaging hack -- a battery-powered free phone booth using 802.11, VoIP and a satellite IP uplink. This was placed in the desert at the Burning Man arts festival deep in the remote Nevada Black Rock playa, exactly where you wouldn't expect a working phone booth to be. With cheap VoIP people were able to call all over the world. The reactions of people to such incongruous technology were great fun and emotional as well. There's a page about the phone including details of building it and live experiences including totally non-gratuitous photos of naked people using technology. (There, that ought to stress-test my new server!)"

Comments

Google Cache

[Anonymous Coward]

http://66.102.7.104/search?q=cache:pj83shgh248J:ww w.templetons.com/pq/playa.html+&hl=en&client=firef ox-a/

Re:Results of new server stress-test:

[grozzie2]

Lots of folks are under the mistaken impression that a /. attack takes out servers. It rarely does that. What it does do, is totally flood the incoming/outgoing network pipe. If your server is on the far side of a t-1 or equivalent connection, the connection doesn't stand a chance, and the episode ends up being just like a distributed syn flood, all the incoming connections, but not enough bandwidth to deliver the responses. *nix boxes tend to survive fine, some flavours of windows boxes will do the bsod in this case, tcpip stack blows buffers in ring0 driver code. OTOH, if you are sitting in a data center with a 100 mbit connection to the upstream router, which has gigabit feeds to the internet, you should have no problem withstanding the onslaught of the /. crowd.

I will admit, on a new server, this is a pretty slick trick to stress test the whole system. Just suggest nudie pics available to the /. crowd, sit back, and watch to see if the upstream routers can deal with the loads. It's a far better way to see if your upstream providers have problems than sitting back and waiting till there's real business/money on the line. I've got a new load balanced cluster going live for a client in a couple weeks, probly gonna steal a page from your book here, I've always known the /. test was a good one, never thought to spice the blurb with the hint of nudie pics.