How Memory QoS Boosts Container Performance in Cloud‑Native Kubernetes
This article explains the memory challenges of containers in cloud‑native Kubernetes environments, describes how Kubernetes maps Request/Limit to cgroups, outlines the limitations of overcommit and mixed workloads, and details Alibaba Cloud Linux 2's Memory QoS features—including memory.min, memory.high, async reclaim, and global water‑mark adjustments—that improve fairness and latency, supported by Redis benchmark results.
