3 ms·
>> But one thing that Tailscale didn't do well (at least early on) is performance. AFAIK currently user space wireguard-go is faster then kernel implementation
by nekonek2 3y ago
>> But one thing that Tailscale didn't do well (at least early on) is performance.
AFAIK currently user space wireguard-go is faster then kernel implementation due to improvements[1] that landed there such as Generic receive offload (GRO) and TCP Segmentation Offload (TSO).
[1] https://tailscale.com/blog/throughput-improvements/ https://tailscale.com/blog/throughput-improvements/