Cloud Concepts and Architecture
What to know
Alibaba Cloud's global infrastructure is organized as regions (geographic areas) containing zones (isolated datacenters with independent power and network). A region is chosen for latency, data-residency/compliance, and service availability; zones within a region are chosen for high availability. Understand the shared responsibility model: the provider secures the physical layer, hypervisor, and managed-service internals; you secure your data, identities, configurations, and guest OS on IaaS. Know the deployment models (public, dedicated/VPC-isolated, hybrid via Express Connect or VPN) and the core service taxonomy: ECS (compute), VPC/SLB (network), OSS/NAS/Disk (storage), RDS (managed database), RAM (identity).
Traps
Questions often test region vs. zone confusion - multi-zone gives fault tolerance within a region, but only multi-region survives a regional outage. Watch for "data must not leave country X" phrasing, which forces a region choice over cost or latency. Another trap: assuming the provider patches your ECS guest OS - it does not. Also distinguish elasticity (automatic scaling to demand) from scalability (capacity to grow), and vertical scaling (bigger instance) from horizontal (more instances).
Rule of thumb
If a scenario mentions compliance or data residency, pick the region first, then use multiple zones for availability. If it mentions surviving a datacenter failure, the answer is multi-zone, not multi-instance on one zone.
Worked example
A workload must stay in-country and survive a single datacenter failure. Choose a region inside the country, deploy across at least two zones, and front instances with SLB - not a single-zone multi-instance setup.