3 ms·
Huh, would be interesting to see some non-tailscale benchmarks of this. Assuming the kernel impl is actually optimized it should be theoretically impossible to
by SOLAR_FIELDS 3y ago
Huh, would be interesting to see some non-tailscale benchmarks of this. Assuming the kernel impl is actually optimized it should be theoretically impossible to exceed the performance with userland wg?
- SparkyMcUnicorn 3y agoI ran the same benchmarks they listed here[0], and did some practical tests. As of a week after the article being written, Tailscale was faster than kernel wireguard. [0] https://tailscale.com/blog/more-throughput/ https://tailscale.com/blog/more-throughput/