High Availability
High availability eliminates single points of failure through redundancy and failover, keeping the service up even when components fail.
Purpose
Eliminate single points of failure, maximizing service uptime.
Mechanism
Redundancy: multi-instance/multi-AZ/multi-region deployment. Failover: health checks automatically deregister failed instances; DNS/LB switches to backups. Stateless services are easy to make HA; stateful ones need replication (primary-replica/consensus). Availability = 1 − MTTR/(MTBF+MTTR).
Use cases
- System design
- Fault tolerance
- Business continuity
Related terms
Service Level Agreement
An SLA is a contractual reliability commitment to customers (e.g., 99.99% availability); breaches trigger compensation, stricter than an SLO. GLOSSARY
CAP Theorem Variants
The CAP theorem states a distributed system can satisfy at most two of consistency, availability, and partition tolerance; its variants refine the trade-offs. GLOSSARY