1. Organizing #
A readable fleet.
Principles #
- Systematic labels at installation:
env,team,region,role. - One configuration per use case (web, databases, Kubernetes), not per server.
- Shared destinations so that URLs and tokens are managed in a single place.
- Gateways to centralize heavy processing and network egress.
2. Reducing and protecting #
Paying for what is useful.
Rules #
- Measure throughput per source before any reduction, then after each processor.
- Filter and deduplicate early; send the full detail to cheap object storage rather than losing it.
- Mask sensitive data in the edge collector, and verify it with a snapshot.
- Route each data type to the destination that justifies it.
3. Deploying #
Without breaking collection.
Rollouts #
- Always a small first batch; progressive rollouts dev, staging then production when the edition allows it.
- Compare the draft with the live version before each rollout.
- Roll back with Roll forward rather than fixing in a hurry.
- Do not delete a resource used by older versions.
Reference diagram and quick check #
Checklist.
Checklist #
- Server in HA or Bindplane Cloud; TLS on OpAMP
- Configurations exported to Git and applied by CI
- Labels, fleets and gateways
- Masking at the source, routing per destination
- Persistent queue on gateways
- Monitoring of collectors, throughput and queues
