Movement replication based on character

Hi,

I am programming replicated movement for a spaceship i am working on. It uses physics simulation and therefore we can’t use the charactermovementcomponent. I am figuring out how the movement component is working and how i can adjust it to my needs. The spaceship is based on a pawn and i am currently just using a static mesh.

Questions:

  1. is it wise to use the physics system for a spaceship in a server/client environment that needs to have replication of position/rotation etc?
  2. If yes, any suggestions how to do the client/server replication, while keeping the server authorative. Keep in mind the speed of the spaceship can be quite high, but has to be accurate without stutters after server corrections.
  3. How to keep this logic performing for say 100 spaceships close by? Is physics simulation even the way to go then?

Questions about the character movement implementation:

  1. What is the idea of the saved moves list? If i understand it correctly the client stores moves and removes them after server acknowledges them. If a server updates comes by the list of current saved moves is replayed. What effect does this have and when should one use this? relevant for my case?
  2. What is the reason that it does not support physics simulation? Too hard?

What i already tested:
Simple replication where both the autonomous client and server simulate the movement keeping the server authoritative. Even with a ping of 30 this already gives stutters sometimes, since server adjustments come later and screw it up. Possibly just smoothing the corrections of the server would already do the job?

I already know of [Video] Player-Controlled Replicating Physics Movement, Simulating Physics! - C++ - Epic Developer Community Forums!
This is not alright for me, since i really need to have an authoritative server.

Thanks!