Hello All,
I love the docker compose approach to self hosting zrok. What are the recommended approach to achieve high availability when the server is done? I am thinking whether there is a way for the access share side to auto switch to a different server when the current used one is down.
Glad to hear it's working for you, @stanworld.
TL;DR you can scale zrok frontends and ziti routers pretty easily to handle huge amounts of traffic, and it's more complicated to achieve true HA for the control planes.
I can sketch out building the Docker instance out with high availability, but you'd probably be the first to do it IRL.
- Ziti router - create Ziti points of presence near your zrok shares and frontends
- Ziti controllers - controller HA is in beta, so you might want to wait on this piece. It uses Raft.
- zrok frontends - they're stateless, so you can put an array of them behind an LB or RR
- zrok controllers - I think you'd have to do active/inactive on this piece for now.
There are mentions of scaling in this doc: Personalized Frontend | Zrok