The AWS Billing Machine and Optimizing Cloud Costs
summary
Stripe loads the AWS cost and usage report into Redshift with a daily ETL (one table per month and a cross-month view), uses the public on‑demand cost for consistent attribution, normalizes instance sizes to an X‑large equivalence, and enforces tags for product ownership. Reserved instances are bought regionally, convertible, no‑upfront, and account for roughly 50% of realized savings; purchases follow automated coverage targets (e.g., 90% per family with a 10% band) and a manual approval step. Engineering fixes include right‑sizing, follow‑the‑sun CI scaling, and single‑AZ Hadoop consolidation that halved costs.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:59Cost and performance alignment
- 4:08Measuring AWS costs at scale
- 12:23Buying reserved instances
- 19:06Automating RI purchases
- 22:12Building cost optimization culture
- 29:21Summary and Q&A
related talks
27mMastering AWS Cost Optimisation
21mHow you can scale up with Amazon EC2 for up to 90% less
49mAWS Cost Reduction - Experiences and Strategies
33mHow we do cloud cost efficiency at Notion
32mBlending Spot and On-Demand Instances and Optimizing Costs
55mOptimizing Costs on AWS Cloud
41mA cloud cost savings spectacular
48mCloud FinOps & Kubernetes Optimisation at Scale