Blending Spot and On-Demand Instances and Optimizing Costs

on YouTube ↗32m

" summary

Shows how to blend Amazon EC2 Spot, On‑Demand and Reserved models with Savings Plans, Auto Scaling (weighted instance pools and capacity-optimised or lowest-cost allocation), SageMaker checkpointing, the Instance Scheduler and Compute Optimizer to cut bills. A customer runs 4532 Spot instances with only 128 terminations (4532 spot instances / 128 terminations) and achieves a 75% cost reduction by falling back to on‑demand for failures and using mixed instance types.

AI-written summary · about

" chapters (10)

1. 0:00Introduction to cost optimization
2. 3:41EC2 pricing models overview
3. 6:57Spot instances customer case study
4. 8:47Savings plans explained
5. 12:00Deep dive into Spot instances
6. 17:57Capacity management strategies
7. 19:23Auto scaling with Spot instances
8. 26:34Compute Optimizer guidance
9. 28:03Workload examples on AWS
10. 30:35Key takeaways and next steps

" related talks

• 21mHow you can scale up with Amazon EC2 for up to 90% lessJoseph Yep · AWS re:Invent 2019
• 42mAWS Spot instances: huge savings (or huge risks?)Taras Postument · code.talks 2019
• 43mSpot InstancesDerek Felska · DevOps World | Jenkins World 2019 San Francisco
• 30mGetting the most out of Amazon EC2Bruno Laurenti · AWS Summit AMER 2020
• 34mSelecting and optimizing Amazon EC2 instancesBoyd McGeachie · AWS AMER Summit May 2021
• 31mEKS Cost Optimization with EC2 Spot and A1 InstancesArthur Petitpierre · AWS Container Day 2019 San Diego
• 05mCutting Cloud Costs in COVID-19!Jeff Smith · DevOpsDays Chicago 2020
• 31mGetting More Out of Amazon EC2Matt Thomson · AWS Summit Online 2020