Cloud Technology and Services

Assuming you know AWS offers many individual services but not how they fit together, by the end of this domain (the largest on the exam at 34%) you can place an unfamiliar requirement onto one map read top to bottom. First you decide how to reach AWS at all (the Management Console, CLI, SDKs, APIs, or Infrastructure as Code). Then you decide where your resources physically live across the global infrastructure of Regions, Availability Zones, and edge locations. Only then do you pick a service category (sometimes called a pillar), starting with the four core ones, compute, storage, database, and networking, and layer higher-level AI/ML, analytics, and application-integration categories on top. The exam tests whether you can place an unfamiliar requirement onto this map, not whether you can operate any one service.

Most Cloud Practitioner service questions are two-step: first name the category ("this is a compute need" / "this is object storage"), then name the service inside it. The four core categories partition cleanly: compute runs code, storage holds data at rest, databases give data structure and query, networking connects and protects. Higher-level categories sit above them: AI/ML services add intelligence without ML expertise, analytics services turn raw data into insight, and application-integration services (queues, topics, event buses) wire components together. Picking the wrong category is the most common avoidable error, so anchor on the data shape or workload type before reaching for a service name.

Across every category, AWS offers a spectrum from self-managed to fully managed, and you generally choose the most managed option that still meets the requirement. In compute this spectrum runs Amazon EC2 → containers → serverless; in databases it runs self-hosted-on-EC2 → Amazon RDS → Aurora and DynamoDB; in analytics it runs cluster-based EMR → serverless Athena and Glue. More management means more control and flexibility but more undifferentiated work; less management means AWS handles patching, scaling, and availability for you. Recognizing this one spectrum explains why two services in the same category coexist.

Resilience in this domain comes from how you place resources, not from any single product. The global infrastructure is one hierarchy: a Region contains multiple Availability Zones, so spreading a workload across multiple AZs is the default high-availability pattern, while using multiple Regions addresses disaster recovery, latency, and data sovereignty. Edge locations (the network behind Amazon CloudFront) are a separate, much larger layer that caches content close to users for performance, apart from the Regions hosting your compute. Knowing which infrastructure layer solves a given non-functional requirement (uptime vs. latency vs. compliance) is itself an exam objective.