When database data and logs live on separate persistent volumes, independent snapshots can capture different moments. A volume group snapshot provides one crash-consistent recovery point across PVCs.
The demo scrubs both volume histories together and marks the same snapshot time. The feature is stable in Kubernetes 1.36; CSI driver support and restore testing still determine recoverability.
When to use
Use it when a stateful workload spanning PVCs needs a coordinated recovery point.