Embracing Karpenter to scale, optimize & upgrade Kubernetes (cmc2025)

Kubernetes is still quite a popular choice with wide community adoption to run containerised workloads in the Cloud, but it doesn’t come with batteries included. And some of that is intentional to allow freedom to make different choices or extend its functionality as needed. For example scaling compute nodes is one of the things which is not built-in. Making sure you’re doing it in most efficient and cost-efficient way is paramount. But it’s not just efficienty than separates Karpenter (an open-source node lifecycle management) from other options, but also how it can help you stay on top with compliance, patching and drift. The project has come a long way in the last couple of year and it was also adopted by CNCF/SIG Autoscaling making it alternative approach compared to de-facto Cluster Autoscaler project. I this talk I’ll show how to set it up, different use cases and demonstrate hands-on what to expect in the real world scenario. Licensed to the public under https://creativecommons.org/licenses/by-sa/4.0/ about this event: https://cfp.cfgmgmtcamp.org/ghent2025/talk/P7CTWQ/