Hot Standby
Definition
Backup system that's fully running and ready to take over instantly if the primary system fails, ensuring minimal downtime and service continuity.
Use Cases
- Netflix: Keeping customer-facing streaming services available during infrastructure failures — Uses an active-active, multi-region architecture on AWS with redundancy across Availability Zones and regions, so traffic can be shifted when a component or zone fails (Improved resilience and reduced customer impact from localized failures by shifting traffic to healthy capacity)
- GitHub: Maintaining availability of core platform data during outages — Uses database replication and failover strategies to keep services running when primary components fail, including standby capacity and operational runbooks for failover (Reduced downtime risk for critical services by enabling faster recovery when primary systems are impaired)
- Capital One: High availability for customer banking applications and supporting data stores — Runs production workloads on AWS using multi-AZ patterns and managed database high-availability features (such as standby replicas) to support rapid failover (Higher service continuity for customer applications and improved recovery posture for infrastructure failures)
Provider Equivalents
- AWS: Amazon RDS Multi-AZ
- Azure: Azure SQL Database Active Geo-Replication
- GCP: Cloud SQL High Availability (HA)
- OCI: Oracle Data Guard
Frequently Asked Questions
- What's the difference between Hot Standby and Cold Standby?
- Hot standby is already running and continuously updated, so it can take over quickly (often seconds or less, depending on the system). Cold standby is powered off or not fully configured; you must start it and restore data before it can take over, which usually takes much longer.
- When should I use Hot Standby?
- Use hot standby when downtime must be very low (tight RTO) and data loss must be minimal (tight RPO), such as payments, trading, healthcare systems, or high-traffic SaaS. It’s also a good fit when you need automated failover and can justify paying for always-on duplicate capacity.
- How much does Hot Standby cost?
- Costs are typically close to running an additional production system: extra compute for the standby, extra storage for replicated data, and network/replication charges (especially across regions). Managed services may bundle standby costs (for example, paying for a second instance in a Multi-AZ setup). The biggest cost driver is keeping a fully provisioned, always-on environment that matches primary capacity.
Category: cloud
Difficulty: intermediate
Related Terms
See Also