Cloud-hosted databases — RDS, Cloud SQL, Azure SQL — typically sit inside a private subnet reachable only through a jump host. Database access from a developer's workstation traditionally required two steps : open an SSH tunnel in a terminal, then connect the database client to the local tunnel port. Forgetting the terminal step left the connection silently broken ; rotating the tunnel after a key change broke every active session.
Database Workbench now captures SSH tunnel parameters directly against the server node in the browser IDE. The connection lifecycle opens the tunnel before the database connection establishes, refreshes the tunnel when the SSH session drops, and tears down the tunnel cleanly on disconnect. Cloud-database access from the IDE no longer requires a parallel terminal ; the operator's mental model collapses to a single connection per server.
- Per-server tunnel metadata. Each server node in the IDE carries its own tunnel configuration — jump-host address, port, username, key reference, local-port allocation strategy.
- Key reference, not key content. The platform's credential store holds the SSH private key ; the server node references it by ID. Keys never live in the browser-side metadata.
- Transparent reconnect. Tunnel drops trigger an automatic re-establish on the next query attempt ; the operator sees a brief reconnection indicator, not a session loss.