In designing Kubernetes, for instance, we document and (mostly) enforce swap off for many of the reasons laid out here. Kubernetes takes over the management of

In designing Kubernetes, for instance, we document and (mostly) enforce swap off for many of the reasons laid out here. Kubernetes takes over the management of host overcommitment, and being able to react correctly to OOM and near-OOM depends to some degree on having a clear understanding of the actual memory use on the system.