Kubernetes Resource Management
Running application in the containers is getting more and more popularity. And with that, it is usually the case that one compute node runs up to several hundreds of applications at the same time. In such setups the problem of "noisy neighbours" arise, with a need to plan and control how much resources each application will be able to consume. In this presentation it is going to be covering current state of Kubernetes to control resources to mitigate "noisy neighbour" issues, talk about resource management in underlying layers (CRI, OCI, Linux kernel, hardware), provide some examples on how to use different extensibility points in Kubernetes stack that can be utilized to inject fine grained resource control.
As a final part of presentation, it will be demo of dynamic container resource management.
Cloud Software Architect
Linux Software Engineer
Alexander has 20+ years of experience in areas of Linux distributions, SCM, Networking, Infrastructure, Release Engineering, Continuous Integration & Delivery, Cloud Orchestration. Previous projects: BlackCat Linux, ASPLinux, OpenWall, Maemo, MeeGo, Tizen, Yocto Currently employed by Intel, Open Source Technology Center in Edge and Cloud Stacks team as Cloud Software Architect, focusing on Kubernetes.
Krisztian has 20+ years of experience writing embedded software and middleware, using Open Source software in various industry segments, including telcos, mobile infra suppliers, and mobile handset manufacturers. His previous projects include Maemo, MeeGo, Tizen, and Yocto. He works for the Intel Open Source Technology Centre in Finland, focusing on Kubernetes and the Cloud Infrastructure.