Integer Programming to optimize costs in Kubernetes

Based on my master’s dissertation, in this topic I will cover an introduction to mathematical optimization and Integer Programming and how these can help reduce pod scheduling costs for us in Kubernetes, building an algorithm that could be a Kube-Scheduler plugin and seek an optimal pod-to-node allocation scenario, always aiming for the cheapest node or a rearrangement of pods to reduce scaled nodes, thus reducing costs for the operated infrastructure.



Camilla Martins

Non-binary. Punk. Docker Captain. Google Developer Expert. I’m a Senior Site Reliability Engineer and currently work at Storyblok, with a postgraduate degree in Computational Forensics and ...