Redundancy
Definition
Having backup copies or duplicate systems that can take over if the primary one fails. Like having a spare tire in your car or a backup power generator.
Use Cases
- Netflix: Keep streaming available during instance, Availability Zone, or service disruptions — Runs services on AWS with redundancy across multiple Availability Zones using load balancing and auto scaling; uses distributed data stores and designs services to tolerate failures. (Improved availability and resilience; reduced customer impact during infrastructure failures.)
- Google: Maintain high availability for global consumer services (e.g., Search, Gmail) despite hardware failures — Uses redundancy at scale across many machines and multiple data centers/regions, with replication and automated failover to keep services running when components fail. (Continuous service operation despite frequent underlying hardware failures; high reliability at global scale.)
- Amazon: Keep e-commerce storefront and checkout available during traffic spikes and component failures — Uses redundant fleets of application servers behind load balancers, multi-AZ database deployments, and replicated storage; capacity scales horizontally to avoid single points of failure. (Higher uptime and better customer experience during peak events; reduced risk from single-system failures.)
Frequently Asked Questions
- What's the difference between redundancy and high availability (HA)?
- Redundancy means having duplicate components (extra servers, replicas, multiple network paths). High availability is the outcome you’re aiming for (minimal downtime). Redundancy is a common way to achieve HA, but HA also depends on failover automation, monitoring, and good architecture.
- When should I use redundancy in cloud architecture?
- Use redundancy when downtime would hurt your business or users, such as customer-facing websites, payment systems, APIs, and critical internal tools. It’s especially important when you have uptime targets (SLAs/SLOs), need maintenance without outages, or must tolerate failures of a VM, disk, or an entire Availability Zone.
- How much does redundancy cost in the cloud?
- Costs usually increase because you run extra capacity and store extra copies of data. Common cost drivers include: additional compute instances (active-active or warm standby), cross-zone or cross-region data replication, extra load balancers, higher database tiers for multi-AZ or replicas, and added network egress for replication. You can control cost by choosing the right redundancy level (single-region multi-AZ vs multi-region), using autoscaling, and right-sizing standby capacity.
Category: cloud
Difficulty: intermediate
Related Terms
See Also