2025-11-08 –, Crystal Dining Room
Your autoscalers are fighting. HPA scales out, VPA scales up, and KEDA drains the queue, a constant war that thrashes resources and kills stability. We will stop the fight, using unified observability and custom business metrics to turn that chaos into a coordinated orchestra, making your system stable, efficient, and finally, intelligent.
This will not be a theoretical talk. We will do this live, with a real Rube Goldberg meme-generator application deployed on a Kubernetes cluster. We will unleash a load test that deliberately creates scaling conflicts, then watch the Grafana dashboards that prove the chaos. Finally, with our unified observability in place, we will run the same test and watch on-screen as HPA, VPA, and KEDA move in perfect concert, adapting to the load gracefully, and proving that the fight is over.
- Attendees will learn how to use the Prometheus Adapter to connect any custom metric to the Kubernetes HPA. This overcomes the limitations of default metrics and enables application-aware autoscaling.
- Attendees will learn why scaling on business-level metrics like P99 latency or error rates is superior to using CPU/memory.This approach creates a more stable and cost-efficient system that scales proactively based on actual user experience, not just infrastructure stress.
- Attendees will learn a framework to orchestrate HPA, VPA, and KEDA so they work together, not against each other. They will leave with the required knowledge to mitigate resource conflicts and build a resilient system where the right autoscaler is always used for the right job.
Nic is an experienced hands-on technologist, evangelist and product owner who has been working in the fields of Cloud-Native technologies, Open Source Software, Virtualization and Datacenter networking for the past 20 years.
Passionate about enabling users and building cool tech solving real-life problems, you'll often see him speaking at global tech conferences and online events, spreading the word and walking the walk with customers and users.