Planning and Implementing a Cloud Solution
What this domain tests
This domain covers translating requirements into architecture: choosing compute, storage, and network options, and sizing them appropriately. Expect scenario questions that give constraints (cost, latency, durability, traffic pattern) and ask for the best fit.
Concepts to master
- Compute choice. Managed/serverless versus self-managed VMs versus containers. Match the workload: bursty or event-driven work favors serverless; steady, lift-and-shift workloads may favor VMs; container orchestration suits microservices.
- Storage classes and durability. Object storage classes trade retrieval cost and availability against storage cost. Know that lifecycle rules move objects between classes and that redundancy options affect durability.
- Networking. Subnets are regional; instances and services live in zones. Understand VPC peering, shared VPC, firewall rules (which are stateful and tag- or service-account-based), and load balancer types by traffic (HTTP(S), TCP, internal versus external).
- High availability. Spread across zones; use managed instance groups with health checks and autoscaling for resilient VM tiers.
Traps to expect
Questions often present a cheaper option that violates a stated durability or latency requirement. Watch for confusion between regional and zonal resources, between firewall rules and routes, and between internal and external load balancing. Another trap: assuming autoscaling fixes a stateful workload without shared storage.
Rule of thumb
Write the constraints down (RTO/RPO, latency, budget, traffic shape), then eliminate options that violate any hard constraint before comparing cost. Worked example: a spiky, event-driven image-resize job with unpredictable traffic is best served by a serverless function triggered by object uploads, not a always-on VM cluster - the VM option is cheaper only if utilization stays high.