4 ms·
> P2P... in different NATs ,,,isn't possible. NAT killed[1] the ability to write true p2p network software. > with UDP hole punching, my tool climbs over NAT
by pdkl95 7y ago
> P2P... in different NATs
,,,isn't possible. NAT killed[1] the ability to write true p2p network software.
> with UDP hole punching, my tool climbs over NAT
First, UDP hole punching is very complicated[2] with reliability dropping fast as the complexity of the NAT(s) increases. It might work "most of the time" when each side is a simple RFC1918
private network behind a router with a well-written stateful firewall+ipmasquerade using a real, publicly routable IPv4 address. I wouldn't assume hole punching works after adding carrier grade NAT, unusual NAT configurations, and/or unfortunately common poorly written routing software.
> Requirement to use
This is missing: "Both sidesz need access to and permission to use a TURN server".
--
[the actual hole-punching method seems fine (I've only skimmed it). I'm only criticizing the claim about offering "P2P communication" when the method still needs an imprimatur from a non-NAT 3rd party.
[1] https://news.ycombinator.com/item?id=20169372 https://news.ycombinator.com/item?id=20169372
[2] https://bford.info/pub/net/p2pnat/ https://bford.info/pub/net/p2pnat/