Resilience engineering
Redundancy, isolation, graceful degradation, and circuit breakers keep partial failures from becoming outages.
Back to the 8 Pillars
Pillar 4
The Reliability Pillar
Build systems that recover gracefully from failure and serve users consistently at scale — with SLOs, error budgets, and resilient architecture.
Resilience
Error budgets
Fast recovery
OVERVIEW
Reliability is availability under failure
Cloud workloads fail. The Reliability pillar designs for failure, measures it, and makes recovery a repeatable practice.
Error budgets
Balance velocity and stability with SLOs and error budgets that tell teams when to ship and when to fix.
Fast recovery
Automated failover, runbooks, incident response, and blameless postmortems reduce mean time to recovery.
CAPABILITIES
What the Reliability pillar covers
From HA/DR design to observability and incident management.
SLOs & error budgets
User-facing reliability targets, measured and defended with error budgets and burn-rate alerts.
HA/DR architecture
Multi-AZ, multi-region, backups, and disaster-recovery plans tested regularly, not just documented.
Observability
Metrics, logs, traces, and health checks that reveal failure modes before users notice them.
Runbooks & chaos testing
Documented response procedures and game-day exercises that validate recovery in realistic conditions.
MATURITY
Three levels of reliability maturity
Move from hoping it stays up to engineering systems that fail safely.
Level 1
Baseline
Backups, basic monitoring, and documented recovery steps exist for production workloads.
Level 2
Standardize
SLOs, runbooks, automated alerting, and tested failover are part of every service launch.
Level 3
Optimize
Chaos engineering, predictive detection, and continuous reliability improvements driven by data.
GET STARTED
Build reliable systems
Read the full Reliability pillar documentation or run your first automated review with WAFPass.
SLOs
HA/DR
Chaos tests