Sharethrough’s adoption for Graviton and Spot
summary
Sharethrough migrates from a third‑party broker to native AWS: ECS capacity providers tied to one ASG per capacity provider, two capacity providers per AZ (Spot primary, On‑Demand standby), five NLBs (one per AZ) fronted by Route 53 weighted routing, and two Lambdas (one for Spot failover, one daily for capacity selection). Results: >99% Spot usage and >90% Graviton adoption, a daily Lambda that picks stable Spot instance types, seamless failover after a 20‑minute alarm, $1M+ yearly savings, and provisioning down from 5 to 3 minutes.
AI-written summary · about
chapters (21)
- 0:00Introduction and speakers
- 0:44What is Sharethrough
- 2:13Scale and AWS partnership
- 3:15Third-party spot solutions
- 5:45Cost motivation for change
- 6:06Native AWS alternatives
- 7:52Architecture design overview
- 10:01Spot and on-demand setup
- 11:07Lambda automation logic
- 14:38Stateless workloads and scaling
- 17:06Capacity providers vs ASGs
- 18:07Route 53 and NLB routing
- 20:20Performance impact and scaling
- 22:09Provisioning time improvements
- 23:32Graviton and instance selection
- 25:18Priority capacity optimization
- 26:26Final results and savings
- 27:13Pricing and free AWS services
- 28:18Alarm timing impact
- 29:23Spot price history and placement
- 30:08Monitoring and dashboards
related talks
42mAWS Spot instances: huge savings (or huge risks?)
47mReliable, secure, and efficient cloud operations
17mWhen Clouds Stop Raining Discounts: Surviving the Drought
31mEKS Cost Optimization with EC2 Spot and A1 Instances
39mHow automating cloud operations accelerated time to market at NU
43mSpot Instances
21mHow you can scale up with Amazon EC2 for up to 90% less
46mArchitecting for Success and Runway