AWS Spot instances: huge savings (or huge risks?)
summary
Claims up to 90% savings on EC2 spot instances, tested running production for more than six months; concrete configs include Auto Scaling Groups with multiple instance types (example m5.large), Spot Fleet / capacity-optimized allocation, and ECS clusters using t2/t3 mediums. Interruptions arrive with a two minute interruption notice, so a Lambda-based drain or ASG lifecycle hooks are used to move containers. Capacity planning recommends 70–80% reserved capacity to absorb evictions. REX-Ray (EBS) is used for stateful workloads; Spot Instance Advisor guides instance selection; one production fleet runs five months without interruption.
AI-written summary · about
chapters (21)
- 0:00AWS Spot instances intro
- 3:37EC2 purchase options
- 5:21Spot instances pricing
- 7:06Spot interruptions risks
- 8:34Spot management tools
- 10:40Auto Scaling with Spot
- 12:10Spot instance pools
- 15:20Spot use cases
- 16:43Spot Advisor
- 18:07CloudFormation challenges
- 21:38ECS with Spot instances
- 23:02ECS capacity planning
- 24:54Handling ECS interruptions
- 26:45Stateful apps on Spot
- 29:15Persistent storage tools
- 33:11Rex-ray setup example
- 34:17Spot examples project
- 36:01Real-world Spot savings
- 38:50When Spot instances are not suitable
- 40:41Spot limitations
- 42:07Q&A
related talks
43mSpot Instances
21mHow you can scale up with Amazon EC2 for up to 90% less
47mReliable, secure, and efficient cloud operations
31mEKS Cost Optimization with EC2 Spot and A1 Instances
32mBlending Spot and On-Demand Instances and Optimizing Costs
31mGetting More Out of Amazon EC2
34mSelecting and optimizing Amazon EC2 instances
05mCutting Cloud Costs in COVID-19!