AWS Karpenter v0.5 Now Generally Available
Today, AWS announced that Karpenter, a new open-source Kubernetes cluster autoscaling project, is now Generally Available with version 0.5 and ready for use in production environments. Karpenter is a flexible, high-performance Kubernetes cluster autoscaler that helps improve application availability and resource utilization. Karpenter launches right-sized EC2 instances in response to changing application load in under a minute. These EC2 instances are based on the specific needs of a cluster’s workloads, such as compute, storage, acceleration, and scheduling requirements. Today, Amazon Elastic Kubernetes Service (EKS) supports clusters using Karpenter on AWS, although Karpenter is designed to work with any conformant Kubernetes cluster.
Today, AWS announced that Karpenter, a new open-source Kubernetes cluster autoscaling project, is now Generally Available with version 0.5 and ready for use in production environments. Karpenter is a flexible, high-performance Kubernetes cluster autoscaler that helps improve application availability and resource utilization. Karpenter launches right-sized EC2 instances in response to changing application load in under a minute. These EC2 instances are based on the specific needs of a cluster’s workloads, such as compute, storage, acceleration, and scheduling requirements. Today, Amazon Elastic Kubernetes Service (EKS) supports clusters using Karpenter on AWS, although Karpenter is designed to work with any conformant Kubernetes cluster.
Kubernetes customers need to continuously adjust the compute capacity of their clusters to support workloads as they scale and to improve cost efficiency. Previously, customers needed to create dozens of EC2 Autoscaling Groups for the Kubernetes Cluster Autoscaler to work as expected to take advantage of the elasticity of the AWS Cloud. This increased operational overhead and degraded performance as their clusters grew. Moreover, customers who needed to provision hundreds of diverse EC2 instances quickly, such as when training machine learning models, experienced expensive scheduling latency, which slowed the pace of their innovation and increased costs.
Karpenter is designed to provision new EC2 instances and schedule Kubernetes pods all in under a minute. Karpenter dynamically chooses the EC2 instance types best suited to what Kubernetes pods need with minimal configuration and no additional AWS infrastructure. As workloads scale, Karpenter automatically adds or removes the instances required, reducing the need for costly over-provisioning and preventing slow, expensive scale-downs. Customers get the capacity their clusters need right when they need it because Karpenter directly integrates with EC2. This means that customers can more easily take advantage of deep discounts from Spot and compute savings plans, further reducing costs.
Learn more and get started with Karpenter today by reading the AWS News launch blog or by visiting karpenter.sh.