Tracks/Capstone & Applied Practice/Module 4.4
Track 4 — Capstone & Applied Practice

Module 4.4: Cloud FinOps & Infrastructure Economics

Cloud cost anatomy, reservation strategy, spot instance economics, FinOps culture, and team-level cloud budgets.

3 Lessons~50 minIntermediate-Advanced
1

Lesson 1: Cloud Cost Anatomy

Cloud spending is the fastest-growing line item in most engineering budgets. Understanding the anatomy of cloud costs — compute, storage, network, and managed services — is the first step to optimization.

Compute (40-60%)

EC2/GCE/Azure VMs, containers, Lambda functions. The largest cost component. Over-provisioning is rampant because engineers fear performance issues.

Average utilization of cloud compute: 20-30%. Opportunity: right-size to 60-70% utilization.
Storage (15-25%)

Block storage, object storage, database storage. Data grows but is rarely cleaned up. 30-50% of stored data hasnt been accessed in 12+ months.

Lifecycle policies can reduce storage costs 40%+ by moving cold data to cheaper tiers.
Network (10-20%)

Data transfer between regions, zones, and to the internet. Often the most surprising cost component because its hard to predict.

Cross-region transfer: $0.01-0.02/GB. At scale: a chatty microservice architecture can generate thousands of dollars in network costs.
📝 Exercise

Pull your last 3 months of cloud bills. Break down by: compute, storage, network, managed services. Which category has the highest waste?

2

Lesson 2: Reservation & Commitment Strategy

On-demand pricing is 3-4x more expensive than committed pricing. A proper reservation strategy is the easiest 30-60% cost reduction available.

Reserved Instances (1yr/3yr)

1-year commitment: 30-40% savings. 3-year commitment: 50-60% savings. Risk: if your needs change, you pay for unused capacity.

Strategy: reserve your baseline (minimum consistent load), use on-demand for peaks.
Savings Plans

AWS Savings Plans / GCP CUDs offer flexibility across instance types. Better for dynamic workloads than traditional RIs.

Cover 60-70% of your baseline with Savings Plans. Leave 30-40% on-demand for flexibility.
Spot/Preemptible Instances

60-90% cheaper than on-demand. Can be terminated with 2 minutes notice. Perfect for batch jobs, CI/CD, and fault-tolerant workloads.

CI/CD on spot: save 70-80% on build costs. Production databases: never use spot.
📝 Exercise

Analyze your cloud usage: what percentage is consistent (reservable) vs. variable (on-demand)? Calculate savings from committing 60% of your baseline to 1-year reservations.

3

Lesson 3: FinOps Team & Culture

Cloud cost optimization is not a one-time project — its an ongoing practice. FinOps creates a culture where engineering teams are accountable for their cloud spend.

Cost Allocation

Tag every resource with team, environment, and project. Allocate 100% of cloud costs to business units. Untagged resources go to a "shame" budget that forces cleanup.

Target: 95%+ of cloud resources properly tagged within 6 months.
Team Cloud Budgets

Give each team a cloud budget. They own it. Overspend: justify or optimize. Underspend: reallocate. This creates natural incentives for efficiency.

Teams with budgets reduce waste 20-40% within the first quarter.
Cost Anomaly Detection

Automated alerts when spending exceeds baseline by 20%+. Catches: forgotten resources, misconfigured auto-scaling, runaway batch jobs.

Set up: daily cost reports + weekly trend analysis + monthly cost review meetings.
📝 Exercise

Implement cost tagging for your top 5 most expensive cloud services. Set up a daily cost anomaly alert (>20% above 7-day average).

← Module 4.3Next: Module 4.5 →
Unlock All — $199/yr