Skip to content

Disaster Recovery & Backups

Hope is not a strategy. We must plan for catastrophic failures (e.g., a regional AWS outage, a dropped production table, or a ransomware attack).

1. Key Objectives

  • RPO (Recovery Point Objective): The maximum amount of data we are willing to lose. Target: 5 minutes.
  • RTO (Recovery Time Objective): How fast the system must be restored and functional. Target: 1 hour.

2. Database Backups

  • Automated Snapshots: The database provider (e.g., RDS, Supabase) must take full automated snapshots daily.
  • Point-in-Time Recovery (PITR): The database must have PITR enabled, allowing us to roll the database back to any exact second within the last 7 days. This is our primary defense against an accidental DROP TABLE or corrupting migration.

3. Infrastructure as Code (IaC)

To meet our RTO, all infrastructure (servers, load balancers, DNS) must be defined in code (e.g., Terraform or Pulumi). In the event of a total regional failure, we must be able to deploy the entire infrastructure stack to a new region via a single CI/CD script.