Skip to content
SENS Platform Docs

Backup and Restore

This document defines initial backup and restore expectations.

Backup and restore are mandatory because the platform stores long-term telemetry and customer data.

  • PostgreSQL relational data.
  • TimescaleDB telemetry data.
  • Audit logs.
  • Export metadata.
  • Configuration needed to restore deployments.
  • Kubernetes secrets where applicable.
  • Helm values for each environment.
  • Container image versions.

Raw payloads:

  • Approximately one month.

Normalized measurements:

  • At least five years.

V1 must document:

  • Backup frequency.
  • Backup storage location.
  • Restore procedure.
  • Restore validation process.
  • Responsible operator.
  • Expected RPO.
  • Expected RTO.

Recovery Point Objective means maximum acceptable data loss.

Initial target: to be decided.

Recovery Time Objective means maximum acceptable restore time.

Initial target: to be decided.

A single Kubernetes server is not a backup strategy.

Persistent volumes can fail.

Database files can be corrupted.

Operator mistakes can delete data.

A separate backup location is required.

A backup is only useful if restore has been tested.

Restore should be tested in test or staging regularly.

  • Backup tooling.
  • Backup storage location.
  • WAL archiving strategy.
  • Full backup frequency.
  • Restore testing frequency.
  • RPO and RTO.