Cloud Architecture
Models and boundaries
Cloud architecture questions usually test whether you can separate overlapping concepts. Know the NIST essential characteristics: on-demand self-service, broad network access, resource pooling and multi-tenancy, rapid elasticity, and measured service. Map IaaS, PaaS, and SaaS to who manages what; the shared responsibility line moves with the service model. Public, private, hybrid, and community deployment models differ in ownership and tenancy, not simply in location.
Resilience vocabulary
Scalability is the ability to grow; elasticity is automatic growth and shrinkage with demand; horizontal scaling adds instances, while vertical scaling adds capacity to one instance. High availability minimizes downtime within a fault domain, often using multiple availability zones and health-checked load balancing. Disaster recovery restores service after a larger event and is driven by RTO (acceptable downtime) and RPO (acceptable data loss). Fault tolerance means continued operation despite component failure; redundancy is the duplicated capacity that makes it possible.
Traps
Candidates mix HA with DR, RTO with RPO, and scaling with elasticity. Multi-AZ is not multi-region. Object, block, and file storage have different access and durability characteristics. A snapshot is not automatically a backup until a restore has been tested.
Worked example
A service needs 99.99% availability. Multi-AZ active-active within one region survives an AZ failure but not a regional outage. If the business cannot accept regional downtime, design multi-region with replicated data and DNS or global load-balancer failover; otherwise document the accepted risk. Rule of thumb: HA keeps running; DR gets running again.