4.4 AWS Cost Management and Monitoring Tools
Key Takeaways
- Cost Explorer visualizes and forecasts spend and produces RI/Savings Plan purchase recommendations — it reports, it does not alert.
- AWS Budgets sends actual and forecasted alerts and can trigger Budget Actions (apply SCPs/IAM policies or stop EC2) when thresholds are crossed.
- Compute Optimizer uses ML on utilization data to recommend right-sized EC2, ASG, EBS, Lambda memory, and Fargate.
- Cost Anomaly Detection learns normal spend with ML and proactively flags spikes — no manual threshold required.
- Trusted Advisor surfaces idle/underutilized resources; only 7 core checks are free, full cost checks need Business or Enterprise support.
Cost Explorer — Visualize and Forecast
AWS Cost Explorer is the analysis surface for billing data. It does not send alerts.
| Capability | Detail |
|---|---|
| Cost breakdown | by service, account, Region, tag, usage type |
| Forecasting | up to a 12-month projection from history |
| Recommendations | optimal RI and Savings Plan purchases |
| Granularity | monthly/daily; hourly + resource-level is opt-in |
Cost Allocation Tags
Tags are how you slice spend by team or project. AWS-generated tags (e.g., aws:createdBy) and user-defined tags (e.g., Department:Engineering) must each be activated in the Billing console before they appear in Cost Explorer.
On the Exam: "Track cost per department" → user-defined cost allocation tags + Cost Explorer. "Forecast next quarter" → Cost Explorer forecasting.
AWS Budgets — Alert and Act
AWS Budgets sets targets and notifies via SNS/email. It distinguishes Actual alerts (cost already exceeded) from Forecasted alerts (projected to exceed). Budget Actions can automatically respond.
| Budget type | Tracks |
|---|---|
| Cost budget | actual + forecasted spend vs. target |
| Usage budget | service usage (e.g., EC2 hours) |
| RI / Savings Plan utilization | are commitments being used? |
| RI / Savings Plan coverage | what % of usage is covered? |
| Budget Action | Effect when threshold hit |
|---|---|
| Apply an SCP | restrict accounts in an OU |
| Apply an IAM policy | deny new resource creation |
| Stop EC2 / RDS | halt running instances |
Compute Optimizer — ML Right-Sizing
| Resource | Recommendation |
|---|---|
| EC2 / Auto Scaling groups | instance type + size |
| EBS volumes | volume type + size |
| Lambda | memory allocation |
| ECS on Fargate | CPU + memory |
It analyzes real CloudWatch utilization, so "right-size based on actual usage with ML" points to Compute Optimizer, while "RI purchase recommendation" points to Cost Explorer.
Cost Anomaly Detection — Proactive ML Alerts
Cost Anomaly Detection learns each monitor's normal spending and flags unusual spikes without any manual threshold, sending SNS alerts with root-cause attribution. It can monitor by AWS service, account, cost allocation tag, or cost category. The exam contrast: Budgets needs a number you define; Anomaly Detection catches the spend you did not anticipate.
Trusted Advisor — Best-Practice Checks
| Category | Example cost checks |
|---|---|
| Cost Optimization | idle/underutilized EC2, idle ELBs, unattached Elastic IPs, low-use EBS |
| Performance | over-utilized instances |
| Security | open security groups, root MFA |
| Fault Tolerance | Multi-AZ RDS, versioning, Route 53 health checks |
| Service Limits | quotas nearing their cap |
| Tier | Checks | Support plan |
|---|---|---|
| Core | 7 checks (S3 bucket perms, security groups, IAM, root MFA, EBS/RDS public snapshots, service limits) | Basic (free) |
| Full | all checks incl. idle resources & RI optimization | Business / Enterprise |
Tool Selection Cheat Sheet
| Need | Tool |
|---|---|
| "Where is money going / forecast it" | Cost Explorer |
| "Alert me when spend exceeds $X" | AWS Budgets |
| "Am I running the right instance size?" | Compute Optimizer |
| "Catch an unexpected spike automatically" | Cost Anomaly Detection |
| "Find idle/underutilized resources" | Trusted Advisor |
| "Cost per business unit" | Cost allocation tags + Cost Explorer |
Worked Example: Reading a Tool Question
The exam rarely names a tool directly; it describes a goal and expects you to map it. Train on the verbs. "Show, analyze, forecast, or recommend a purchase" → Cost Explorer. "Alert when spend crosses a number I choose" or "stop resources at a limit" → AWS Budgets with Budget Actions. "Right-size based on actual utilization using ML" → Compute Optimizer. "Detect an unexpected spike I never anticipated" → Cost Anomaly Detection. "Find idle load balancers, unattached IPs, underutilized EC2" → Trusted Advisor. "Attribute spend to a team or project" → cost allocation tags surfaced in Cost Explorer.
A frequent distractor pits Budgets against Cost Anomaly Detection. The deciding word is whether a threshold is known: if the question gives a dollar figure ("alert at $5,000"), it is Budgets; if it says "unexpected" or "without setting a threshold," it is Anomaly Detection. Another distractor pits Cost Explorer RI recommendations against Compute Optimizer. Cost Explorer recommends what to buy (RIs/Savings Plans) from billing data; Compute Optimizer recommends what to run (instance type/size, Lambda memory) from utilization metrics.
Organization-Wide Cost Governance
For multi-account estates, AWS Organizations with consolidated billing pools usage so volume tiers and RI/Savings Plan discounts apply across all accounts, and a single payer account receives one bill. Service Control Policies (SCPs) can prevent expensive actions — for example, denying launches of oversized instance families or restricting Regions — as a preventive cost control rather than a reactive alert. Cost Categories let you group accounts and charges into business dimensions (by team, environment, or product) that then flow into Cost Explorer, Budgets, and Anomaly Detection.
On the Exam: "Centralize billing and maximize volume/commitment discounts across many accounts" → AWS Organizations consolidated billing. "Prevent teams from launching costly resource types" → SCPs. "Automatically stop or restrict resources when a budget is hit" → Budget Actions.
Cost-Tool Trap Table
| Tempting wrong answer | Why it fails |
|---|---|
| Cost Explorer to send a spend alert | It visualizes and forecasts but does not alert |
| Budgets to catch an unforeseen spike | Budgets needs a threshold you define in advance |
| Compute Optimizer for RI purchase advice | That is Cost Explorer's recommendation engine |
| Trusted Advisor full cost checks on free tier | Only 7 core checks are free; rest need Business+ |
| Separate accounts solely to tag-track cost | Cost allocation tags do this without account sprawl |
A finance team wants an automatic notification as soon as the projected end-of-month AWS bill is expected to exceed a set spending limit. Which service should they configure?
Which AWS service uses machine learning on utilization metrics to recommend optimal EC2 instance types and sizes?
Which TWO services help identify underutilized or idle AWS resources? (Select TWO)
Select all that apply
A company's AWS bill jumped 40% over the past week with no planned changes. Which service would have PROACTIVELY alerted them to this without any threshold being set in advance?
Which Trusted Advisor check is available on the Basic (free) support plan?