Storage Architecture
OSC provides regional block and object storage from a distributed storage platform operated across the Availability Zones of a region.
Block storage is used for persistent volumes and worker-machine disks. Kubernetes workloads request it through standard storage classes and PersistentVolumeClaims. The underlying volume is created and attached automatically; customers do not manage the storage nodes or storage cluster directly.
Object storage provides an S3-compatible interface for applications that store unstructured data, backups or artifacts. It is consumed independently from the Kubernetes node lifecycle and is suitable for data that should not be tied to a single worker machine.
The storage platform is shared across customers but uses logical separation, access controls and encryption mechanisms to isolate data. Storage nodes are dedicated to the regional service and are not used as customer compute nodes.
The distributed design protects against individual disk and node failures and is built to use multiple zones. Even so, storage replication is not the same as application backup.
Customers should define retention, backup and restore procedures according to the business value of their data.
Applications should select the storage type that matches their access pattern.
Block storage is appropriate for filesystems and databases that expect a mounted device.
Object storage is better suited to large collections of immutable or API-addressable objects.