Storage that outlives the pod or container that uses it, the bridge between ephemeral compute and persistent state.
A volume in Kubernetes is storage attached to a pod, with two main flavors: ephemeral volumes that share the pod's lifecycle (emptyDir, configMap), and persistent volumes (PV) that outlive the pod and are bound by a PersistentVolumeClaim (PVC). PVs are typically backed by cloud block storage (EBS, GCP PD, Azure Disk) or networked file systems (EFS, FSx, NFS). For stateful workloads, picking the right access mode (ReadWriteOnce, ReadOnlyMany, ReadWriteMany) is a foundational architectural decision.
Stateful workloads on Kubernetes are subtle: a pod may be rescheduled to a different node where the original PV's zone-locked storage isn't available, or a PVC may be deleted (with its data) when a Helm chart is uninstalled. Storage class defaults around reclaim policy (Retain vs Delete) have caused multiple high-profile data-loss incidents. Treat persistent volumes with the same care as databases.
See the part of the platform that handles volume (persistent volume) in production.