Short version: people who made circuit-switch telephone networks had a radically different view of the future Internet than the packet-switching young rebels. For a while it wasn’t clear what path we were going to take. Every now and then I revisit this article to remember what it was like when we didn’t know.
At one point lots of people were convinced that we'd build everything on top of ATM with its 53 byte payloads. The unknown future was weird.
I started briefly reading about ATM in the past months.
Nothing much and mostly superficial stuff, but it’s probably one of the more bizarre network protocols I’ve seen.
I’ve been meaning to pick up some ATM switches and other equipment to see if I can toy with it and create my own little network. I already did this with ISDN.
“A camel is a horse designed by committee.” ATM was a do-everything solution made to meet a bunch of mutually incompatible requirements like “must have minimal latency for voice, so the packet size must be tiny” and “must be capable of fast data streams, so act like a circuit where routing is trivial”.
I highly approve of your experiment! When you’re done, you should have a good appreciation of the full WTF of its design.
At one point lots of people were convinced that we'd build everything on top of ATM with its 53 byte payloads. The unknown future was weird.