LEARN. PRACTICE. GROW.Understand the concept
A region is a geographical area; Availability Zones provide isolation within a region. Distributing components across zones can reduce local failure impact, but the application must support that design. Select regions using data requirements, latency, service availability, and operations, not map proximity alone.
Apply and decide
High availability and disaster recovery have related but different objectives. Replication may propagate corruption; backups should enable recovery of an appropriate state. Multiple instances do not remove a single dependency. Examine the entire chain and demonstrate recovery against agreed RTO and RPO.
Workplace application
For a service with three web instances and one database, draw their shared dependency. Discuss what happens if the database fails or receives a logical deletion. Define with the business how long recovery may take and which data loss is tolerable. Request restoration evidence matching those objectives. One tier’s availability does not demonstrate whole-chain availability, and replication may propagate an error.
Two instances in different zones still depend on one database without a recovery strategy. The design must address that dependency.
Common pitfalls
Confusing local redundancy with complete service recovery.
Related topics: Choose compute and data services · Objects, blocks, files, and networking
Redundancy in one component does not prove end-to-end resilience.
Reference: AWS global infrastructure · CLF-C02