Malte Schwarzkopf A Q&A with Malte Schwarzkopf, co-author of the Omega paper. tl;dr: Malte explains how the primary goal of Google Omega’s shared-state design was flexibility in software engineering, rather than scalability. He explains that shared-state scheduling enables useful policies, such as priority preemption and interference-aware scheduling. Prior two-level scheduler designs (e.g. in YAR
![Omega, and what it means for Kubernetes: a Q&A about cluster scheduling – Kismatic](https://cdn-ak-scissors.b.st-hatena.com/image/square/9aad5e783a60c435b48010b2e8b3c3a7b836b64d/height=288;version=1;width=512/http%3A%2F%2Fblog.kismatic.com%2Fwp-content%2Fuploads%2F2015%2F06%2Fmugshot-2012.jpg)