Unlocking Kubernetes IO Insights with ACK’s New Storage Monitoring Dashboards

Background

With the rapid adoption of containerized applications, growing workloads, and increasing complexity, real‑time, precise monitoring of storage performance and health has become essential. Alibaba Cloud Container Service for Kubernetes (ACK) introduces enhanced storage monitoring to give users deeper insight into both internal and external storage resources.

New Dashboard Optimization Approach

The update adds dedicated monitoring panels for different storage types, improving existing dashboards and introducing three new ones for cloud disk, NAS, and OSS volumes. It also enriches cluster, node, and pod dashboards with storage‑related charts.

Supported Storage Types

Local storage (e.g., hostPath, emptyDir)

Secret and ConfigMap (metadata storage, low observability requirements)

PVC‑based storage, with recommended Alibaba Cloud options: cloud disk, NAS, OSS

New Dashboard Content

Internal Cluster Storage

RootFS monitoring – usage ratio and real‑time read/write rates

Pod temporary storage – usage and inode consumption

External Cloud Storage

Cloud Disk – summary (name, namespace, usage), real‑time read/write speed, latency, throughput

NAS – same metrics as cloud disk, with mount point address displayed in PVC tables

OSS – summary (bucket name, namespace, usage), read/write throughput, operations per second (OSS and POSIX), hotspot file statistics

Dashboard Screenshots

Each storage type has a dedicated monitoring panel showing the metrics described above. The PVC information tables list PV/PVC names, namespaces, nodes, device names, and usage percentages. Additional charts display overall space consumption and real‑time I/O data.

Cluster, Node, and Pod Monitoring Enhancements

Cluster dashboards now include a PVC overview table. Node dashboards show PVC summaries and read/write data. Pod dashboards add RootFS and Ephemeral Storage metrics, covering:

All non‑emptyDir temporary volumes

Pod log files stored on the node

Writable layers of all containers in the pod

Enhanced Ephemeral Storage Monitoring

The Ephemeral Storage Usage(%) chart only displays containers that have resources.limits.ephemeral-storage set. RootFS metrics are exposed by cAdvisor; however, when the runtime is containerd, cAdvisor does not provide pod‑level metrics. A regression caused a previous fix to be reverted, but ACK’s csi‑plugin (v1.28.3‑eb95171‑aliyun or later) restores this capability.

Conclusion

The upgraded ACK storage monitoring gives users comprehensive, fine‑grained visibility into storage details across the cluster, helping quickly locate IO bottlenecks and ensure stable application performance.

References

Resource Management for Pods and Containers – https://kubernetes.io/docs/concepts/configuration/manage-resources-containers/

Monitoring cAdvisor with Prometheus – https://github.com/google/cadvisor/blob/master/docs/storage/prometheus.md

cAdvisor pull request fixing containerd issue – https://github.com/google/cadvisor/pull/2872

cAdvisor pull request rollback – https://github.com/google/cadvisor/pull/2964/files#diff-68b17ffbed25b18140ee93332bd2a1f259a4eff46ccf93ca800bd2386041437b