TL;DR We just released Pomerium v0.31 with a new file-based databroker backend. It eliminates the operational overhead of managing a separate data persistence layer by introducing a self-healing, infrastructure-agnostic storage mode that recovers in milliseconds and keeps sessions alive.
Scale and Performance Without the Overhead
Few teams have a full-time DBA. And most Pomerium deployments don’t need the control, scale, and complexity of Postgres.
Pomerium’s new file-based databroker backend removes the burden of managing a separate data persistence layer without sacrificing performance.
**Meet the Self-Healin…
TL;DR We just released Pomerium v0.31 with a new file-based databroker backend. It eliminates the operational overhead of managing a separate data persistence layer by introducing a self-healing, infrastructure-agnostic storage mode that recovers in milliseconds and keeps sessions alive.
Scale and Performance Without the Overhead
Few teams have a full-time DBA. And most Pomerium deployments don’t need the control, scale, and complexity of Postgres.
Pomerium’s new file-based databroker backend removes the burden of managing a separate data persistence layer without sacrificing performance.
Meet the Self-Healing File-Based Databroker
The new backend embeds Pebble, the storage engine behind CockRoachDB — a fast, embedded key-value store built for production workloads. It’s designed to cover 80% of real deployments without the overhead of Postgres.
What you get:
- Self-healing recovery. Raft handles clustering and leader election. If one node fails, another takes over in less than a second.
- Infra-agnostic. Run anywhere — Kubernetes, VMs, bare metal, or air-gapped networks. No managed database required.
- Persistent and safe. Sessions and data survive service restarts.
- Built for scale. Pomerium’s databroker is optimized for Pebble’s key-value store.
- Observability ready. Exposes metrics and traces for direct visibility in Grafana or your preferred tool.
How Self-Healing Works with Optional Clustered Mode
Teams can decide between clustered and non-clustered mode. We recommend clustered mode for teams required self-healing with automatic leader election in the event of a failure.
When cluster mode is active:
- The leader writes new data.
- Followers replicate changes in real time.
- If the leader fails, Raft based leader election promotes a follower to leader immediately.
- The new leader already has the full state and resumes writes.
No volume reattachment, or manual replay. Just automatic recovery.
Quick Start Example
Enable file-based clustering with two simple lines:
databroker_storage_type: file
databroker_storage_connection_string: file:///var/pomerium/databroker
That’s it. Your sessions persist. Your data survives restarts. Failover happens automatically.
The Right Default for Most Teams
For most deployments, the file-based databroker is the recommended choice: simple, self-healing, and ready for production.
Pomerium now handles storage the way operators always wanted — quietly, reliably, and without additional operational overhead.
Try Pomerium v0.31
Upgrade to the latest version or deploy fresh today.
Request a demo or connect with your Pomerium team to see how v0.31 simplifies your setup.
NB: Raft is available with Pomerium Core and Enterprise.
Photo by Glen Carrie on Unsplash