Current problem
I'm running a self-hosted instance via Docker Compose on a single machine, currently working well for my community's size. I'm planning ahead for growth and want to understand the supported path for scaling beyond a single host before I actually need it — but I can't find clear, authoritative documentation on this. Searches turn up conflicting information (different sources describe different databases, container counts, and deployment approaches), so I'm not confident which guidance actually applies to the current version of Fluxer.
Proposed change
A documentation page (or an answer here I can act on) covering:
- Whether there's an official/supported architecture for distributing services across multiple machines, versus just vertically scaling one host.
- Whether Cassandra is the current recommended database for horizontal scaling versus Postgres for standard deployments — and if so, guidance for migrating or running them side by side during a transition.
- Which component tends to become the bottleneck first for a growing community (database, realtime gateway/websocket layer, media/file storage, or voice via LiveKit) — something to help operators know what to monitor for as an early warning sign.
- Whether Docker Swarm, Kubernetes/Helm, or manually networked separate Compose stacks is the intended approach for anyone who does need to go multi-node.
Additional information
Currently running the standard Docker Compose deployment (Postgres, Valkey, NATS, Meilisearch, SeaweedFS, LiveKit, Caddy, plus the API/Gateway/Worker/shard services) on a single Windows Server host via WSL2. Happy to share more detail about the current setup if it's useful context. Not an urgent issue — planning ahead rather than hitting a wall today.