FinOps for Engineers: Why Your Architecture Has a Price Tag
summary
Cloud architecture decisions map directly to invoices: using premium block blob for 10 TB/month across four regions jumped to roughly 80,000/month, a restore of 50 GB cost 1.2 million NOK because 150k files triggered per-object archive charges, and naive autoscaling on CPU causes unexpected VM spins. Engineering changes that work include Azure Storage Inventory report plus lifecycle policies to move cold data, compress before archive to avoid per-object costs, and use queue length for autoscaling instead of CPU; these cut storage by ~30% and other costs by ~20%.
AI-written summary · about
chapters (11)
- 0:00FinOps for engineers
- 2:39Hidden costs in cloud
- 6:32Billing mirrors design
- 9:06FinOps engineering dimension
- 11:14Autoscaling bill inflation
- 18:04Storage lifecycle traps
- 25:40Logs budget killer
- 28:34Microservices network costs
- 32:18FinOps in workflows
- 36:43FinOps non-functional requirement
- 42:31Key takeaways
related talks
46m10 Strategies to keep your Cloud costs under control
41mFinOps and Cloud Cost Optimization
17mBoosting Cost Efficiency by Cost Aware Architecture
45mCloud Computing: Cloud in a cost-effective way
30mAnd the CFO Wept: AWS Cost Control
15mWell-Architected Framework: Cost Optimization
46mAzure on the cheap
31mGetting the Most out of Every Dollar Spent in Cloud