Auto-Scaling, Watch My Instances Grow and Shrink
summary
An open source auto-scaling service lives in the Cloud Foundry incubator, shipping an MVP that uses memory-based MVP rules only, exposes control via the service broker API, and stores state as it migrates from CouchDB to Postgres migration. Policies declare min/max instances, metric rules, schedules and cooldowns; actions are triggered by continuous feedback loops and by scheduled windows. Custom metrics and router response-time metrics are planned, vertical scaling and no CPU support are explicitly excluded for now.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:28Cloud Foundry Incubator
- 2:50What is Autoscaling
- 3:49Scaling based on metrics
- 5:07Dynamic Scaling
- 8:22Advantages and Drawbacks
- 9:18Scheduling
- 10:52Bandhas Service
- 13:32Repeating Policy
- 14:17circling Policy
- 14:51Whats Next
- 20:22Rules
- 28:12Overcommit
related talks
19mThe Taming of the Queue
33mA Tale of Two System Crashes: Innovation and Accidental Chaos
36mCustomize Your Cloud Foundry UI through Stratos Extension Framework
31mDeploy Your First Cloud Foundry App
33mHow dorma+kaba Leverages and Deploys on CloudFoundry
18mAPIs at Scale Through Minimalism in Code and Architecture
34mScaling Cloud Development Environments on K8s: Enterprise-Scale Adoption Lessons
20mSystems @Scale Summer Live Q&A Session #2