4 ms·

When I looked into it, webtorrent still requires a centralized server to at the very least set up the peer to peer connection. The problem is that there isnt a

When I looked into it, webtorrent still requires a centralized server to at the very least set up the peer to peer connection. The problem is that there isnt a way to avoid this with WebRTC. It also serves content to the initial peer. How is this serverless? Am I missing something?

callesgg 10y ago
I am with you.
d33 10y ago
https://developer.mozilla.org/en-US/docs/Web/Guide/API/WebRTC/Peer-to-peer_communications_with_WebRTC https://developer.mozilla.org/en-US/docs/Web/Guide/API/WebRT...
johnhenry 10y ago
While the infohash is stored in the blockchain, the other immediate issues is that to get the infohash, you still need to connect to a centralized third-party such as blockexplorer.com. Browsers don't natively support TCP, which is how normal torrent and blockchain clients connect to each other. WebRTC is a workaround for torrents, but there doesn't seem to be a workaround for blockchains If we want true serverless architecture within browsers, we, at least, need browsers that support TCP connections.
- roblabla 10y ago
  What features of TCP would allow serverless connections ? AFAIK, even BitTorrent (DHT) and others connect to a centralized "super-node" which then gives them other nodes to connect to.
  - lgats 10y ago
    If you look long enough you can find an open TCP connection to another client who has the infohash you are looking for—wirhout ever getting the other clients ip from a central server.
  - johnhenry 10y ago
    Sorry, that was hastily written and poorly worded -- more accurately, we would AT LEAST, need browsers that support TCP connections in order to connect to the p2p networks that we already have. In practice we would need a lot more, as even apps like BitTorrent and blockchains still rely on our centralized internet infrastructure to some degree. You mention the BitTorrent DHT, but I don't believe that it's necessary to connect to a centralized super-node. You do, however; need the address of at least one other peer to start. Unfortunately, the only way to assign an address is through our ISP system which is centralized. For true peer-to-peer architecture, normal people would have to connect the internet at the same level that ISPs such as Comcast and AT&T connect to each other... think some sort of large scale mesh network.
murtnowski 10y ago
WebRTC has no built in peer discovery to bootstrap the startup. At minimum a signaling server is will be needed for exchanging descriptions. You may also need to exchange ICE candidate information. So while this isn't serverless, it could be really great for reducing bandwidth costs until you consider that this design does not play well with mobile devices which have usage caps.

callesgg 10y ago

I am with you.

d33 10y ago

https://developer.mozilla.org/en-US/docs/Web/Guide/API/WebRTC/Peer-to-peer_communications_with_WebRTC https://developer.mozilla.org/en-US/docs/Web/Guide/API/WebRT...

johnhenry 10y ago

While the infohash is stored in the blockchain, the other immediate issues is that to get the infohash, you still need to connect to a centralized third-party such as blockexplorer.com. Browsers don't natively support TCP, which is how normal torrent and blockchain clients connect to each other. WebRTC is a workaround for torrents, but there doesn't seem to be a workaround for blockchains If we want true serverless architecture within browsers, we, at least, need browsers that support TCP connections.

roblabla 10y ago
What features of TCP would allow serverless connections ? AFAIK, even BitTorrent (DHT) and others connect to a centralized "super-node" which then gives them other nodes to connect to.
- lgats 10y ago
  If you look long enough you can find an open TCP connection to another client who has the infohash you are looking for—wirhout ever getting the other clients ip from a central server.
- johnhenry 10y ago
  Sorry, that was hastily written and poorly worded -- more accurately, we would AT LEAST, need browsers that support TCP connections in order to connect to the p2p networks that we already have. In practice we would need a lot more, as even apps like BitTorrent and blockchains still rely on our centralized internet infrastructure to some degree. You mention the BitTorrent DHT, but I don't believe that it's necessary to connect to a centralized super-node. You do, however; need the address of at least one other peer to start. Unfortunately, the only way to assign an address is through our ISP system which is centralized. For true peer-to-peer architecture, normal people would have to connect the internet at the same level that ISPs such as Comcast and AT&T connect to each other... think some sort of large scale mesh network.

murtnowski 10y ago

WebRTC has no built in peer discovery to bootstrap the startup. At minimum a signaling server is will be needed for exchanging descriptions. You may also need to exchange ICE candidate information. So while this isn't serverless, it could be really great for reducing bandwidth costs until you consider that this design does not play well with mobile devices which have usage caps.