Design Cost-Optimized Architectures is its own scored domain, and cost trade-offs appear inside resilience and performance questions too. The exam wants you to pick the cheapest option that still meets the requirement — not the cheapest option outright.

Understanding EC2 purchasing models alone can swing a bill by 70% or more, so this is high-leverage knowledge.

EC2 has four core pricing models. On-Demand — pay per second, no commitment, best for short/unpredictable workloads. Reserved Instances — 1 or 3-year commitment for a large discount on steady-state usage. Savings Plans — commit to an hourly spend for flexible discounts across instance families and even Fargate/Lambda. Spot Instances — up to ~90% off spare capacity, but AWS can reclaim them with two minutes' notice, so they suit fault-tolerant, interruptible work like batch jobs.

Right-sizing means matching instance type/size to actual utilization instead of over-provisioning. For storage, S3 tiering (lifecycle rules, Intelligent-Tiering) moves cold data to cheaper classes.

Cost tooling includes Cost Explorer (visualize and forecast spend), Budgets (alerts on thresholds), and the Cost and Usage Report (granular billing data).

A steady 24/7 production database runs year-round on EC2. A separate nightly analytics batch job can restart if interrupted. How do you optimize each?

For the steady workload, buy Reserved Instances or a Savings Plan — predictable, always-on usage earns the biggest committed-use discount. For the interruptible batch job, use Spot Instances at up to ~90% off, since a reclaim just reschedules the work.

The trap answer is using Spot for the always-on database — a reclaim would cause an outage. Match the pricing model to the workload's tolerance for interruption and its predictability.

Cost optimization overlaps every domain. Auto Scaling saves money by removing idle capacity (compute + cost). S3 lifecycle rules and Intelligent-Tiering cut storage bills (storage + cost). VPC endpoints avoid NAT data-processing charges (networking + cost). ElastiCache reduces expensive database load (databases + cost).

The recurring exam lesson: the cost-optimized answer is usually the one that meets the requirement with the least always-on, over-provisioned, or on-demand spend — using commitments, tiering, and elasticity.