OpenObservability Talks Cloud Native Spotlight: KEDA & Dapr Projects - OpenObservability Talks S6E04
23 snips
Sep 28, 2025 
In this discussion, Yaron Schneider, co-creator of Dapr and KEDA, shares insights from his extensive experience in cloud architecture. He highlights how Dapr provides easy abstractions for developers across cloud platforms and the critical role KEDA plays in event-driven autoscaling. Yaron further explores the importance of observability with tracing over logs in distributed systems. Additionally, he touches on the intersection of AI and Dapr, emphasizing innovative durability features for automated agents. A must-listen for enthusiasts of cloud-native tech!
AI Snips
Chapters
Transcript
Episode notes
Origins Of KEDA And Dapr
- KEDA started inside Microsoft to solve autoscaling beyond CPU/memory for event-driven workloads like Kafka and SQS.
- Dapr followed to give application developers a runtime toolbox to consume APIs without vendor lock-in.
Autoscale On Events, Not Just CPU
- Kubernetes default autoscaling (CPU/memory) often reacts too late for queue- or event-driven systems.
- Autoscaling on backlog or custom metrics provides proactive scaling and avoids downstream failures.
Use Prometheus Metrics For Custom Scaling
- Emit Prometheus-compatible metrics from your app if you want KEDA to scale on custom internal signals.
- Use KEDA's Prometheus scaler to turn those app metrics into autoscaling triggers.