Cloud Computing Fundamentals
Core concepts
Alibaba Cloud, like any public cloud, delivers compute, storage, and network resources as metered services. Understand the shared responsibility model: the provider secures the physical infrastructure and virtualization layer, while the customer secures guest OS, applications, data, and identity. Know the service models - IaaS (ECS, VPC), PaaS (ApsaraDB, Function Compute), SaaS - and deployment models (public, private, hybrid). Familiarity with regions and zones is essential: a region is a geographic area, zones are isolated datacenters within it. Resources generally cannot span regions without explicit replication, but can span zones for high availability.
Elasticity and billing
Distinguish vertical scaling (resize an instance) from horizontal scaling (add instances via Auto Scaling). Understand pay-as-you-go, subscription, and preemptible/spot billing, and when each is appropriate. Questions often test cost-awareness: steady workloads favor subscription, spiky or fault-tolerant workloads favor pay-as-you-go or preemptible instances.
Traps to watch
- Confusing a zone failure with a region failure; multi-zone ≠ multi-region.
- Assuming the provider secures customer data - it does not.
- Mixing up elasticity (automatic, demand-driven) with scalability (capacity to grow).
- Forgetting that cross-region replication requires deliberate configuration and incurs latency and cost.
Worked rule of thumb
If a workload must survive a single datacenter outage, deploy across multiple zones in one region. If it must survive a regional disaster or serve users with low latency on another continent, add cross-region replication and a global traffic mechanism. Match the resilience goal to the scope of failure you are defending against.
Exam framing
Read each scenario for the failure domain and the cost constraint; those two signals usually determine the correct architectural answer.