Infinity Is Not a Strategy: Right‑Sizing the Cloud
summary
Argues the “cloud is infinite” myth and gives concrete replacements: use ARIMA forecasts and LSTM-style continuous load models, pick overcommit K from utilization distributions via the C+K overbooking formula, and size buffers with napkin math (example: 118K request buffer to hit a 99% lead‑time target). Reuse industry tactics—N-1 contingency rules, mixed‑integer crew/flow scheduling for real‑time failover, explicit safety stock and automatic load‑shedding—and shows these techniques are built and operationally effective.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 3:30Infinite scale fallacy
- 8:47Right-sizing capacity
- 10:33Airlines capacity lessons
- 13:18Demand forecasting
- 15:26Crew rotation optimization
- 17:52Power grid resilience
- 21:04Real-time load forecasting
- 23:10Logistics forecasting
- 25:17Safety stock policies
- 28:50Redundancy for critical systems
- 30:36Optimizing for scarcity
- 31:39Summary
related talks
05mSponsored Keynote: The Cloud (And Your $$$) is Not Infinite: Dynamic Scheduling at Every Layer
36mSustainable cloud infrastructure: from ideation to deployment- 36mSustainable cloud infrastructure: from ideation to deployment
29mCapacity Constraints Unveiled: Navigating Cloud Scaling Realities
33mThe Rent Is Too High! Cutting Costs on Cloud Real Estate
09mon What to Avoid in Public Clouds
19mFrom discovery to optimization: Four steps to cloud adoption
37mHybrid OnPrem + Cloud Architectures