I like the ideas in this, and that it's got changes that allow it to work over stream based protocols, which means webrtc and tor. However DHTs (Kademlia at least) aren't efficient on a large scale if they're not built on packet oriented protocols. I don't know what the solution is, maybe allowing TCP is just a compromise you have to make sometimes.
Hi, I wrote this post, and this is something I was wondering about the whole time. It seemed like TCP was a necessary compromise, but I wasn't sure how painful the overhead would be. If there are any notable examples you know of where this went wrong (or right) I'd be interested.
If it helps, you may find that when you scale your DHT to 10000+ nodes you will start to have issues with building on TCP. Of course many DHTs never get to that scale. IPFS is an example of one that sits just above that mark and can't figure out why things don't work well.
For comparison, the BitTorrent DHT has 20 MILLION nodes and works fantastically.
This is true. Last time I looked at the DHT in IPFS it was full of kludges and workarounds for performance issues due to the fact that it uses TCP connections underneath.
