Single-node Overview
Single-node hosted is the cheaper, simpler of the two hosted availability tiers. Its control plane — the Hub — runs as a single instance on one compute node, keeping state in an embedded SQLite database on local or single-volume storage. There is no external database to provision, secure, back up, or pay for.
Use this page to understand what the tier gives you and whether it is the right target before you follow the Hub Setup guide.
What this tier provides
Section titled “What this tier provides”- A networked Hub. Unlike Workstation mode (loopback only), a single-node Hub is reachable over the network, so a small team — or you, from multiple machines — can share projects, agents, and state.
- Single- or multi-user. Tenancy is orthogonal to the tier: a single-node Hub can serve one user with a simple token, or many users through an OAuth identity provider with Groups and access policies.
- Low cost and operational simplicity. One VM or one Cloud Run instance backed by SQLite — nothing else to run.
The sqlite database driver (SCION_SERVER_DATABASE_DRIVER=sqlite) is what pins the Hub to a
single instance: a single DB connection and in-memory lifecycle-hook deduplication.
The trade-off: non-HA
Section titled “The trade-off: non-HA”Single-node is not highly available. In exchange for its low cost, it accepts:
- Restart/redeploy downtime. There is only one Hub instance; while it restarts, the control plane is unavailable.
- Single-volume durability. State lives on one volume. Protecting it is your responsibility (snapshots/backups of the SQLite volume).
“Single-node” scopes the control plane only — agents themselves may run on other nodes via runtime brokers.
When to use it
Section titled “When to use it”Choose Single-node hosted when: “A cheap, simple, shared Hub for me or a small team.”
- You want a networked Hub without standing up Postgres, object storage, and load-balanced replicas.
- Occasional restart downtime is acceptable.
- You want the lowest-friction, lowest-cost path to a shared deployment.
If you need to survive node loss and redeploys without downtime, step up to HA hosted instead.
How it differs from HA hosted
Section titled “How it differs from HA hosted”| Single-node hosted | HA hosted | |
|---|---|---|
| Control plane | One Hub instance on one node | Hub replicated behind a load balancer |
| Database driver | Embedded sqlite | External postgres (Cloud SQL) |
| State & durability | Single-volume; non-HA | External DB + object storage; highly available |
| Downtime on restart/redeploy | Yes | No |
| Typical realization | One VM (starter-hub scripts) or one Cloud Run instance + SQLite | Cloud Run (min-instances ≥ 2) + Cloud SQL |
| Cost & complexity | Low | Higher |
| Tenancy | Single- or multi-user | Single- or multi-user |
The two tiers are distinguished purely by the availability tier dimension — see Choosing a Mode for the full picture.
Next steps
Section titled “Next steps”- Hub Setup — configure and run the Hub.
- Deploy on a VM (GCE) — the starter-hub path.
- Auth & Tenancy — single- vs multi-user access.
- Connecting to a Hub — the user-facing journey.