08-16, 15:00–15:45 (Europe/Berlin), Milliways
Language: English
THIS TALK WILL BE HELD IN "RESCUE" MODE DUE TO PRESENTER'S HARDWARE FAILING
Building your own hardware to work with 10 Gigabit Ethernet… without access to measurement or simulation equipment. Can the high frequency electronics deities be appeased with some sacrifices?
Born from building and tearing down a bunch of fibre for CCC event networks: a pocket tool to help quickly troubleshoot our 10GE connections, and get us out of random "square peg, round hole" problems. Passive single-port adapters that can read diagnostic information from optical transceivers are common and easy, but how about something with 4 ports and active circuitry?
Well, electrons don't really like being pushed around at GHz frequencies, start clinging to edges and corners, and turn around and go back the way they came when that's the easier path… HF circuit design has a well-earned aura of respect. One of the well-known reference books is in fact titled "High Speed Digital Design: A Handbook of Black Magic" and assumes you have a fancy (= expensive) electronics lab with nice gear.
Let's, uh, just ignore all that and see the result of building high-speed (10GE - 10.3125Gbaud, 5.16GHz) electronics with just basic microcontroller experience!
NB: this talk will be incredibly boring (or even annoying?) to anyone with actual HF design experience. And I cannot teach you how to do it correctly either 😅
equi's natural habitat at camp is the NOC tent, though he can be a bit skittish when approached with too much gusto. Travelled birdwatchers might also spot him being a maintainer on FRRouting or at IETF and RIPE events. Suitable bait for sniping shots can be found in naked high-speed networking hardware, FPGAs and other embedded hardware.