Loading the context…
Loading the context…
A distinct deployment area within a cloud region, intended to separate some infrastructure failure risks.
An application can place resources in multiple zones so that one zone's failure does not automatically remove every copy. That requires usable capacity, replicated data and a mechanism to redirect work. Merely having another server somewhere is insufficient if it lacks current data or depends on the failed zone. Zones reduce some shared physical risks, but software mistakes or a regional dependency can still affect several together.
Suppose a hypothetical service runs in zones A and B, each normally handling 50 requests per second. Demand totals 100. If A stops, B needs capacity for all 100 requests and access to the required data to continue normally. If B can handle only 60, having two zones has preserved part of the service but not full capacity. The design must account for the failure condition, rather than only normal operation.
Yandex reports Vladimir data-centre shutdown as taxi and cloud services are disrupted