Financial Resiliency Engineering: Taming Cloud Costs
summary
Shopify narrows cloud spend by turning billing into traceable profiles, exporting billing and service data into flame-graph billing profiles (speedscope/prep) to spot that compute engine and GPU drive costs, then applying fixes: VPA recommender to right‑size requests, GKE cluster autoscaler’s optimize‑utilization profile, Terraform policy checks via Atlantis + Conftest, and tagging schema for attribution. Results include reclaiming massive kube-unallocated capacity and decommissioning two hold clusters per region, materially lowering monthly spend.
AI-written summary · about
chapters (9)
- 0:00Introduction to Shopify
- 1:26The challenge of cloud cost
- 4:16Initial optimization efforts
- 9:26Visualizing costs with data
- 13:55Bin packing and utilization
- 18:36Right-sizing workloads
- 23:54Establishing guardrails
- 26:34Future goals and metrics
- 34:44Conclusion
related talks
41mFinOps and Cloud Cost Optimization
47mCost optimization
05mFive Minute Cost-Optimisation
05mCutting 2/3rds of your cloud costs with common sense
48mCloud FinOps & Kubernetes Optimisation at Scale
31mGetting the Most out of Every Dollar Spent in Cloud
33mThe elephant in the cloud: Strategies of cloud cost optimisation
46m10 Strategies to keep your Cloud costs under control