How we do cloud cost efficiency at Notion
summary
A small cloud infra lead identifies an August 2022 AWS spike caused by an EMR autoscaling misconfig running ~20x idle instances, fixes it and helps save millions, then builds a biweekly program that pushes reserved capacity from ~50–60% to reserved capacity to ~100% using compute savings plans and RIs, adds CUR exports into Snowflake for tagging, and catches an S3 abuse case that cuts AWS by 10% S3 bill cut; decisions follow a 20x ROI cutoff for engineering time.
AI-written summary · about
chapters (8)
- 0:00Introduction and Background
- 2:36The Beginning of Notion's Cloud Cost Efficiency Journey
- 4:36First Lessons and Wins in Cost Efficiency
- 9:44Building a Legitimate Cost Efficiency Program
- 13:06Defining Worthwhile Investments and Goals
- 21:33Continuing the Momentum and Future Goals
- 23:00Key Learnings and Conclusion
- 23:59Q&A Session
related talks
41mA cloud cost savings spectacular
21mHow you can scale up with Amazon EC2 for up to 90% less
31mThe AWS Billing Machine and Optimizing Cloud Costs
34mOptimizing Cloud Efficiency with Prompt Engineering
05mCutting Cloud Costs in COVID-19!
18mThe nonstop way to find AWS-recommended savings opportunities
56mSaving on AWS? If not, what are you waiting for?
41mFinOps and Cloud Cost Optimization