Demystifying Kubernetes Resource Management: Everything You Always Wanted to Know … But Were Afraid to Ask

Embark on a journey to unravel the world of Kubernetes resource management, shedding light on foundational primitives and uncovering best practices and real-world scenarios.

We'll touch on how requests and limits artfully abstract the potent capabilities of the Linux kernel while playing a pivotal role in Kubernetes scheduling.

We'll delve into the Quality of Service (QoS) and how Kubernetes determine utilization and act upon node pressure.

We will explain the importance of Resource Quotas and Limit Ranges and lessons learned in multi-tenancy on large Kubernetes deployments.

Finally, we will briefly touch on advanced resource management primitives, such as static CPU manager, Control Topology Management Policies, and NUMA-Aware Memory Manager.

The "Demystifying Kubernetes Resource Management" presentation will significantly benefit the practitioners as it sheds light on important but often overlooked Kubernetes Resource Management primitives.

The audience will understand how Kubernetes works behind the scenes on pod placement, eviction under node pressure, and Linux task scheduling.

We will distill some of the complex discussions happening in the community about the pros and cons of pod limits in simple terms and with a pragmatic view from someone who deployed and managed Kubernetes at scale in a highly regulated environment.

We will bring to awareness the newer resource management features, such as integration with cgroups V2, CPU Manager, and NUMA-Aware Memory Manager.

The presentation will allow the audience to improve their Kubernetes Resource Management skills and ultimately enhance their overall productivity, with a clear call to action to avoid performance degradation, avoid outages, provide better multi-tenancy, and ultimately reduce cloud costs.