AZ-900 Describe cloud concepts Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  The ability to automatically scale resources during peak usage

  Why it's wrong here

  Automatically scaling resources during peak usage is the definition of elasticity, a hallmark of cloud computing that adjusts capacity to match current demand. While elastic scaling improves performance and cost efficiency, it addresses workload fluctuations, not component failures. In contrast, fault tolerance demands that the system survives a hardware or software fault without downtime; scaling merely adds more instances, but a failure in one of those instances could still disrupt service if no redundancy or failover mechanism exists.

• ✓

  The ability to continue operating correctly despite component failures

  Why this is correct

  Fault tolerance is the architectural property that lets a system continue functioning within specification even when one or more of its components fail. This is achieved through redundancy—duplicating critical components (servers, networks, power supplies) so that a single failure can be absorbed by a standby or parallel unit, preventing an outage. A truly fault-tolerant system detects the failure and performs a seamless failover without interrupting user requests, unlike merely rebooting or recovering after the fact.

• ✗

  The ability to restore data after a major disaster

  Why it's wrong here

  Restoring data after a major disaster describes disaster recovery (DR), which is the process of rebuilding systems and data from backups or replicated copies in a separate site after a catastrophic event like a data center loss. DR is focused on recovery point objectives (RPO) and recovery time objectives (RTO), meaning it accepts downtime while data is restored, whereas fault tolerance ensures zero downtime and no data loss during a component failure. DR is a reactive survival strategy, but fault tolerance is preventive and happens automatically as part of the system's runtime design.

• ✗

  The ability to deploy applications in multiple geographic regions

  Why it's wrong here

  Deploying applications across multiple geographic regions is a geodistribution strategy typically used to reduce latency for global users or to provide location-level resilience, such as surviving an entire region outage. It does not, however, define fault tolerance, which is about handling internal component failures like a failed disk, network card, or process within a single deployment. A multi-region setup can improve availability, but unless each region is itself internally fault-tolerant with redundant hardware and software, component failures inside a region can still cause downtime or degraded operation.

Go deeper

Related to this question