Clarify the use case / benefit of relay and discovery servers

I think the documentation does a good job explaining the purpose/benefit of using a relay and discovery on the client side. But it’s less clear to me what the use cases for the relay and discovery servers are.

I’m guessing that running a relay server is about contributing to the community and helping to make relaying overall more robust? But maybe I’m basing that on my associations with the term “relay” as used in Tor. And running your own discovery server — is faster? improves privacy? I’m not sure.

I think the documentation could use just a couple extra sentences somewhere to explain this, but I thought I’d post my question here instead of in a GitHub Issue, in case I missed something.

I’m also interested because I’m helping with the Fedora packages of Syncthing, and I want to make sure the package descriptions are accurate.

Relay server to contribute to the community or because you need one and your own is faster. Discovery server if you need one because you must do global discovery inside your routed, NAT-ed network (unusual) or because you need global discovery but absolutely don’t trust the regular servers for some reason (privacy / paranoia). I expect more people to run relays than discos :slight_smile: