Running Low-Latency Workloads on Kubernetes

Explore strategies for optimizing low latency workloads on Kubernetes in this informative conference talk. Dive into the intricacies of SpiceDB's massively parallel architecture and learn how to fine-tune Kubernetes defaults for improved performance. Discover the power of taints, tolerations, requests, and limits in resource management. Examine the pros and cons of context switching, and gain insights into implementing static CPU manager policies. Enhance your understanding of Kubernetes optimization techniques to achieve optimal performance for latency-sensitive applications.

Intro
Jimmy Zelinskie
SpiceDB is Massively Parallel
Kubernetes Defaults: are you even trying?
Taints & Tolerations
Requests & Limits
Pros
Context Switching
Static CPU Manager Policy