Custom App Autoscaler Using Cloud Controller API - Christopher Decelles
summary
Builds a custom autoscaler that polls REST endpoints returning JSON and makes real scale decisions via the Cloud Controller API (or Java client), packaged as a service broker and installable tile. A worker polls every 30 seconds for rules that compare JSON fields (actuator or custom metrics, firehose reader) and then sets app instances (demo moves 1→2→3 and back based on min/max and equals/greater-than rules). It deploys a worker per service-instance and uses Cloud Controller API and Java client.
AI-written summary · about
chapters (8)
- 0:00Introduction to project
- 1:14Defining scaling rules
- 3:30Leveraging custom metrics
- 6:20System architecture overview
- 8:57Demonstrating data sources
- 12:56UI and tile configuration
- 15:27Scaling demonstration
- 20:05Implementation details and Q&A
related talks
19mThe Taming of the Queue
26mCloud Foundry AutoScaler Service: Features, Design and Roadmap
36mDrive Higher Utilization of Apps in Azure Spring Cloud with Autoscale
16mAutoscaling Distributed System with BOSH
36mCluster Autoscaler Evolution
38m50 Shades of K8s Autoscaling - Mario Leander Reimer
22mIntelligent Autoscaling in Enterprise...
36mAutoscaling Mesos Cluster