Cloud Concepts and Architecture
What to know
Candidates must be fluent in the shared responsibility model, the difference between availability zones (isolated failure domains within a region) and regions (geographic groupings), and the elasticity/vertical-vs-horizontal scaling distinction. Understand service models (IaaS, PaaS, SaaS) by what the customer manages, and deployment models (public, private, hybrid, multi-cloud). The Well-Architected pillars - operational excellence, security, reliability, performance efficiency, cost optimization, sustainability - are conceptual anchors, not trivia. Know that high availability (redundancy within a region) differs from disaster recovery (cross-region, defined by RPO/RTO).
Traps
Questions love to blur "fault tolerance" with "high availability": fault tolerance means zero interruption via redundancy (active-active), while HA accepts brief failover. Another common trap is confusing scalability (capacity grows with load) with elasticity (capacity shrinks automatically when load drops). Watch for options that describe managed services as removing all customer responsibility - customers always retain data, identity, and access configuration duties. Also distinguish vertical scaling (bigger instance) from horizontal (more instances); only horizontal usually implies fault tolerance.
Worked example
A team needs a workload to survive an entire data-center loss with near-zero data loss. A single-AZ multi-instance design fails the "data-center loss" clause; the correct answer is multi-AZ deployment plus cross-region replication for the recovery tier. Rule of thumb: match the failure domain named in the question to the redundancy scope in the answer - if the scenario says "zone," choose zonal redundancy; if it says "region," choose regional replication.