2 ms·
This has been known forever. Stream-based packet tunneling absolutely does outperform packet wrapping in good network conditions, for the simple reason that sm
by ajross 9y ago
This has been known forever. Stream-based packet tunneling absolutely does outperform packet wrapping in good network conditions, for the simple reason that small packets can be combined by the transport layer and avoid the overhead.
The problem is that in the presence of any packet loss at all, every packet lost causes a stall of the whole tunnel until it gets retransmitted, which even in the best case recovery conditions requires two (three? I forget my SACK details) round trips. If one side is using traditional TCP retransmit timers it can be much, much longer.
- aexaey 9y agoWell-known unpredictable TCP-over-TCP performance would only explain OpenVPN/TCP results. But OP has posed also OpenVPN/UDP performance that is 2x...3x slower than SSH port forwarding, so I'd guess there is more to this story (see also andmarios's impromtu test results above).
- Buge 9y agoYour downside I don't think applies in this case. Because it's a single application wanting to make a single reliable connection. So if there was any packet loss at all it would have to stall the whole connection anyways regardless of whether it's wrapped in SSH, OpenVPN, or nothing.
- jacob019 9y agoI've been using a VPN over TCP ssh for a year. The end point is across the world. I get excellent bandwidth and no stalls. Latency is less than anticipated. The thoretical arguments against TCP/TCP do not match my experience.
- ajross 9y agoYeah, real world network paths, especially inter-data-center ones, generally do operate at exactly zero packet loss. Until they don't, alas.