Usage based cost optimisation at scale
summary
National Australia Bank builds a data-driven power management platform that uses VPC flow logs ingested via Data Firehose into S3, an IP-to-application map from CloudTrail, hourly Lambda processing and queries with Amazon Athena to surface inactivity windows, then calculates per-second EC2 cost and generates shutdown schedules. Prototype scales to 500 applications across 2,000+ non-production environments, produces measurable cost savings, and is available as an open-source repo to run in your account.
AI-written summary · about
chapters (9)
- 0:00Introduction and speakers
- 1:11Problem statement and goals
- 3:25NAB's cost optimization background
- 7:03Power management platform limitations
- 9:34Intelligent power management architecture
- 17:44Prototype demonstration and results
- 20:14Adoption and process changes
- 22:45Measurable impact and future plans
- 24:59Next steps and open-source tool
related talks
21mBest Practices for Optimizing Your Cloud Spend with AI
48mCloud FinOps & Kubernetes Optimisation at Scale
49mAWS Cost Reduction - Experiences and Strategies
55mOptimizing Costs on AWS Cloud
41mFinOps and Cloud Cost Optimization
47mCost optimization
24mOptimizing Cloud Costs with Machine Learning - A CloudVane Story
20mAtomics Habits in cost optimizations in the cloud