DataTerminology.comNEW
Also known as: HA/DR
Practices that ensure service resilience and rapid recovery by designing for redundancy, failover, and recovery objectives across failure scenarios.
DESCRIPTION
High Availability (HA) keeps services operational through component failures via redundancy, replication, and failover. Disaster Recovery (DR) restores services after major incidents using backups, cross-region replicas, and runbooks. Together, HA/DR plans define RTO (time to restore) and RPO (data loss tolerance). Strategies span active-active, active-passive, cold/warm/hot standby, and geo-distribution. Testing with game days and chaos experiments validates readiness. Balance cost, complexity, and risk appetite when selecting patterns.
ALTERNATIVE DEFINITIONS
Microsoft Azure Architecture Center
High availability and disaster recovery (BCDR) strategies help keep applications running and data safe during planned and unplanned outages.
Positions HA/DR within broader business continuity and resiliency.
AWS
Disaster recovery (DR) is the process of preparing for and recovering from a disaster.
Vendor definition focusing on the recovery process component.
Google Cloud
Disaster recovery is the set of policies and procedures to enable the recovery or continuation of vital technology infrastructure and systems following a disaster.
Highlights policies and procedures enabling recovery or continuation.
RELATED TERMS
Durability (ACID) Replication Failover Recovery Point Objective Recovery Time Objective Geo-Redundancy Active-Active Active-Passive Chaos Engineering
COMPETING TERMS
Backup and Restore Standby
RELATIONS
- Broader: Disaster Recovery
KNOWLEDGE AREAS
Infrastructure Data Storage & Operations Cloud Data Security MLOps Observability & Telemetry
TARGET AUDIENCE
Cloud Architect DevOps Engineer Data Engineer Data Architect Chief Technology Officer MLOps Engineer Chief Information Officer
EXTERNAL SOURCES
- Wikipedia - High availability
- Microsoft - Resiliency and BCDR overview
- AWS - DR whitepaper and patterns
- Google Cloud - DR scenarios and planning
TOP QUESTIONS
What is High Availability and Disaster Recovery (HA/DR)?
HA/DR combines redundancy to keep services running with procedures to recover from major incidents. It defines acceptable downtime and data loss. Plans must be tested.
Category: Definition
What are RTO and RPO in HA/DR?
RTO is time to restore service; RPO is acceptable data loss. They guide architecture and cost. Tighter objectives require more investment.
Category: Example
What is a common mistake in HA/DR planning?
Unvalidated runbooks and untested failover paths. Regular drills and chaos experiments are crucial. Documentation must be current and actionable.
Category: Mistakes
How do active-active and active-passive differ?
Active-active serves traffic from multiple sites, improving resilience and latency. Active-passive keeps a standby ready for failover. Costs and complexity differ significantly.
Category: Example
What risks exist without HA/DR?
Outages cause revenue loss and reputational damage. Recovery may be slow, with data loss exceeding tolerance. Regulatory penalties may apply.
Category: Risks
How does backup and restore fit into HA/DR?
Backups underpin DR but alone may yield long RTO/RPO. Combine with replication and failover to meet tighter objectives. Test restores frequently.
Category: Requirements
What is a hot topic in HA/DR today?
Multi-region architectures, managed disaster recovery services, and automated failover validation. SRE practices increasingly shape resilience.
Category: Topic
How do I choose an HA/DR strategy?
Start from business impact, RTO/RPO, and regulatory needs. Evaluate patterns (cold, warm, hot, active-active). Balance cost and complexity.
Category: Management
How is HA/DR applied to ML systems?
Replicate feature stores, model registries, and inference endpoints. Ensure deterministic rollbacks and shadow deployments. Monitor model availability and drift.
Category: Example