Cloud Regions and Zones

A cloud resource must still run somewhere physical. The location choice becomes visible when users are far away, a data center fails, or policy requires data to remain in a particular place.

The Geographic Hierarchy

A region is an independent geographic area in which a provider operates infrastructure. A zone is a deployment area within a region. Providers may also offer multi-region placements that replicate or serve data across several regions.

provider footprint
└── region
    ├── zone A
    ├── zone B
    └── zone C

The names are organizational abstractions, not promises that all services have identical failure behavior. Each service documents what a zone or region means for that resource.

Placement Changes System Behavior

Placement Useful property Remaining risk or cost
One zone Simple and often inexpensive A zonal disruption can stop the workload
Several zones in one region Better isolation from one-zone failures A regional disruption still affects every replica
Several regions Wider failure isolation and proximity to distant users More coordination, replication, and transfer cost

Distance also contributes to remote-call latency. Putting compute near users can shorten requests, while putting compute near its data can avoid repeated long-distance calls. Those goals may conflict.

A Placement Check

Choose placement by asking:

  1. Where are the users and dependent systems?
  2. Which failures must the service survive?
  3. Where is the data allowed to reside?
  4. Does the application tolerate asynchronous replication or stale reads?
  5. What extra transfer and operational cost is justified?

A multi-region label is not a complete resilience plan. The application must still route around failures, handle ambiguous network outcomes, and verify that its data service provides the required replication semantics.