Event-driven autoscaling for Serverless Java

on YouTube ↗56m

" summary

Shows how to replace CPU-based HPA with event-driven autoscaling by combining KEDA and Knative for Quarkus Java apps consuming Apache Kafka. Builds a demo pipeline that generates 5 million Kafka messages, deploys a Quarkus consumer, and proves KEDA triggers zero→one quickly while delegating 1→N scaling to HPA so pods reach up to the configured max (demo hit 10). Works faster than HPA alone (scale-up in ~6–10s), but notes cold starts, partition rebalance and single‑pod work imbalance when scaled from zero.

AI-written summary · about

" chapters (10)

1. 0:00Introduction and speaker intro
2. 1:32Kubernetes autoscaling fundamentals
3. 5:08Introducing KEDA for event-driven scaling
4. 9:26KEDA vs KNative for serverless
5. 11:35Demo 1: HPA autoscaling with Quarkus
6. 25:30Demo 2: KEDA autoscaling with Kafka
7. 30:53KNative serverless Java applications
8. 44:20Demo 3: KEDA-KNative integration
9. 50:00Key takeaways and future roadmap
10. 51:47Q&A session

" related talks

• 49mEvent-driven autoscaling for Serverless JavaDaniel Oh · DevWorld 2024
• 52mEvent-Driven Autoscaling for Serverless JavaDaniel Oh · JCON Europe 2023
• 55mEvent-Driven Autoscaling for Serverless JavaDaniel Oh · Devnexus 2023
• 58mEvent driven Autoscaling for Serverless JavaDaniel Oh · Great International Developer Summit 2023
• 35mKEDA - Real Time And Serverless Scaling In KubernetesZbynek Roubalik, Jeff Hollan · KubeCon + CloudNativeCon North America 2022
• 37mKubernetes Event-driven Autoscaling with KEDAZbynek Roubalik, Jorge Turrado · KubeCon + CloudNativeCon Europe 2022
• 34mAutoscaling Made Easy with Kubernetes Event-Driven Autoscaling (KEDA)Tom Kerkhove · Global Azure Virtual 2020
• 31mServerless Java in Action: Cloud Agnostic Design Patterns and TipsDaniel Oh · DevConf.US 2024