unknown

Pomerium v0.31 replaces its Postgres-based storage backend with a custom file-based solution using Pebble (LSM-tree storage) and Raft consensus for clustering. The change addresses operational challenges with managed Postgres instances at scale, including performance issues, configuration complexity, and lack of DBA expertise. The new approach uses Raft for leader election rather than full consistency, forwarding writes to a single leader while followers replicate state for failover. This provides easier management, comparable performance, and automatic recovery while relaxing consistency guarantees since lost data can be rebuilt from external sources.

Sometimes Postgres isn’t the Answer

Zero Trust community exploring secure access for modern infrastructure. Share Pomerium insights, implementation guides, identity-based patterns, or anything related to context-aware security, proxies, and practical approaches that protect systems.

Pomerium

<p>TL;DR: Pomerium v0.31 introduces a new file-based Databroker backend using Pebble and Raft—offering clustered failover, fast reads, and easier ops. Postgres remains fully supported, but for teams without DBAs or facing scale pains, the file-based option strikes a pragmatic balance between performance and simplicity.</p>
