In this session, I'll be discussing the limitations of using Cluster Autoscaler, its advantages, and how Karpenter addresses those limitations.

I’ll also provide a brief overview of EKS cluster architecture, explain how Karpenter operates, and demonstrate how it optimizes cluster nodes automatically.

While Karpenter excels in scaling, its capabilities extend far beyond that to include cost optimization through features like consolidation.

I'll showcase a live demo with an application to illustrate how nodes scale up and down based on pod scheduling.

This demo will highlight how Karpenter optimizes costs by consolidating nodes and selecting the appropriate instance types from various options configured in the provisioner.

Additionally, I’ll demonstrate the significant business impact of Karpenter and explore its use cases beyond typical web/app applications, including its role in data analytics, machine learning, and building foundational models for generative AI.