The author discusses the importance of fair benchmark testing and why they are publishing their results of mesh VPN offerings. They provide guidelines for meaningful benchmark testing like being objective, buying hardware, detuning hardware, testing multiple streams, comparing functionally equivalent things, leveling the playing field, never co-mingling software on a host, tuning everything, and having fun sarcastically. The author then compares the mesh VPN options Nebula, Netmaker, Tailscale, and ZeroTier in three primary tests focusing on transmit, receive, and bidirectional performance. They highlight unique performance characteristics, memory usage, and CPU efficiency of each option.
https://www.defined.net/blog/nebula-is-not-the-fastest-mesh-vpn/