DevOps Cost Optimisation: Cut Cloud Spend 40% Through Engineering Practices

Engineering Owns Cloud Costs

Cloud costs are an engineering problem. Instance sizes, retention periods, replication factors — these decisions determine spend. FinOps only works when engineering is engaged.

1. Infracost in CI/CD

Add cost estimates to every PR. Engineers see monthly cost impact before merging. One line in GitHub Actions shows: 'This change will cost +$347/month'.

2. Environment Scheduling

Stop non-production at 7pm, restart at 8am, shutdown weekends. AWS Instance Scheduler or Azure Auto-Shutdown. Result: 65% reduction in non-prod compute costs.

3. Container Right-Sizing

Kubernetes VPA recommendation mode: most containers over-request CPU/RAM by 3-5x. Implementing recommendations reduces required node capacity by 30-40%.

4. S3 Lifecycle Rules

IA after 30 days, Archive after 90 days, delete after 365. Reduces storage costs 40-60% for log-heavy applications.

5. Spot for CI/CD

GitHub Actions with EC2 Spot runners: 70% cheaper than On-Demand for builds.

Savings Summary

Make cloud costs visible in PRs and dashboards. Cost optimisation becomes intrinsic when engineers see the impact of their decisions.