Your Autoscalers Are Fighting. Here's How to End the Thrash.
Ask your cluster how big checkout-api should be and how many replicas it needs. HPA says 11. KEDA says 24. VPA says the pods are over-provisioned. Karpenter is buying nodes nobody uses, and Argo CD wants it all reset to Git. Each controller does its job. Together, they thrash. Join Sedai engineer Nikhil Kurup to see why, and how to end it.
Register for the webinar
December 16, 1:00-1:45 p.m. (EST)
Attend the webinar to learn:
Why Your Autoscalers Can't Agree
Sizing a workload is one equation. Kubernetes splits it across five controllers that never see each other. VPA moves the denominator HPA divides by. KEDA and a chart's HPA disable each other. Argo CD reverts scale-ups at peak. Karpenter buys hardware for every inflated request. The tools aren't broken. Nobody owns the decision.


Your Autoscalers Are Fighting. Here's How to End the Thrash.
Nikhil Kurup is Senior Vice President of Engineering at Sedai. As a leading expert in building machine learning systems that operate at massive scale, Nikhil ensures that Sedai can safely manage and optimize the cloud at some of the world’s largest companies. He has more than two decades of engineering experience, most recently at H2O.ai, Teradata, and Kore.ai. Nikhil earned a Master's degree in Information and Data Science from UC Berkeley.
Nikhil Kurup
SVP Engineering | Sedai
Stop Refereeing Your Autoscalers.
Start Deciding.