according to: https://github.com/longhorn/longhorn/wiki/Performance-Benchmark it should be a lot closer to device speeds