Reliability
- The question it answers
- Can the workload perform its intended function correctly and consistently, and recover from failure?
- Design principles
- Automatically recover from failure — monitor KPIs and trigger automated remediation.
- Test recovery procedures, not just functionality.
- Scale horizontally to increase aggregate availability — many small resources, not one big one.
- Stop guessing capacity — let demand drive it.
- Manage change through automation.
- Typical services
- Multi-AZ deployments, AWS Auto Scaling, load balancers, Amazon Route 53 health checks, AWS Backup, AWS Elastic Disaster Recovery, Amazon Simple Queue Service (SQS) for decoupling.
- Key terms
- RTO — how long recovery may take. RPO — how much data may be lost.
- Distinguish from
- Performance Efficiency — is it fast and efficient? Reliability is does it stay up and recover?