Menu Close

Chapter 5. Persistent Storage

Container storage by default is not persistent. For example, if a new container build occurs then data is lost because the storage is non-persistent or if a container terminates then of the all changes to its local filesystem are lost. OpenShift Container Platform offers many different types of persistent storage to avoid those situations. Persistent storage ensures that data that should persist between builds and container migrations is available.


For more information about the available storage options in OpenShift Container Platform see Types of Persistent Volumes

When choosing a persistent storage backend ensure that the backend supports the scaling, speed, and redundancy that the project requires. This reference architecture will focus on cloud provider specific storage.


This reference architecture is emerging and components like Container-Native Storage (CNS), and Container-Ready Storage(CRS) will be described in future revisions.

5.1. Persistent Volumes

Container storage is defined by the concept of persistent volumes (pv) which are OpenShift Container Platform objects that allow for storage to be defined and then used by pods for data persistence. Requesting of persistent volumes is done by using a persistent volume claim (pvc) object. This claim, when successfully fulfilled by the system will also mount the persistent storage to a specific directory within a pod or multiple pods. This directory is referred to as the mountPath and facilitated using a concept known as bind-mount.


For more information about the persistent volumes and its lifecycle, see Lifecycle of a Volume and Claim

Persistent volumes can be preprovisioned by the OpenShift Container Platform administrator by creating them in the underlying infrastructure and in OpenShift Container Platform manually, or the administrator can configure OpenShift Container Platform to create automatically the proper persistent volumes when users request them using the dynamic provisioning and storage classes capabilities of OpenShift Container Platform.

5.2. Storage Classes

The StorageClass resource object describes and classifies different types of storage that can be requested, as well as provides a means for passing parameters to the backend for dynamically provisioned storage on demand. StorageClass objects can also serve as a management mechanism for controlling different levels of storage and access to the storage. Cluster Administrators (cluster-admin) or Storage Administrators (storage-admin) define and create the StorageClass objects that users can use without needing any intimate knowledge about the underlying storage volume sources. Because of this the naming of the storage class defined in the StorageClass object should be useful in understanding the type of storage it maps to (ie., HDD vs SDD or Premium_LRS vs Standard_LRS).

5.3. Cloud Provider Specific Storage

Cloud provider specific storage is storage that is provided from Microsoft Azure. This type of storage is presented as an Data disk VHD and can be mounted by one pod at a time. It is needed to configure OpenShift Container Platform with Microsoft Azure settings like the resourceGroup or subscriptionID in the /etc/azure/azure.conf file on masters and nodes as well as in the OpenShift Container Platform masters and nodes configuration file to be able to use VHD as persistent storage for pods. The settings needed are automatically configured as part of the installation process using the code provided in the openshift-ansible-contrib git repository.


For more information about the required settings, see Configuring Azure

Cloud provider storage can be created manually and assigned as a persistent volume or a persistent volume can be created dynamically using a StorageClass object. Note that VHD storage can only use the access mode of Read-Write-Once (RWO).

The VHDs used in Microsoft Azure are .vhd files stored as page blobs in a standard or premium storage account in Microsoft Azure where standard delivers cost-effective storage and premium delivers high-performance, low-latency storage.

5.3.1. Creating a Storage Class

When requesting cloud provider specific storage in Microsoft Azure for OpenShift Container Platform, there are two options to define a storage class:

  • Create a service account in the same resource group where the OpenShift Container Platform cluster has been deployed in Microsoft Azure where all the VHDs will be created.
  • Provide a skuName and location to OpenShift Container Platform where all storage accounts associated with the resource group are searched to find one that matches.

In this reference architecture the first options has been chosen as it is simpler and avoids searching for matching service accounts where they will be provided before using them.

Besides the /etc/azure/azure.conf configuration file, it is required to create a storage account per storage class created in OpenShift Container Platform in order to be able to use dynamic provisioning of volumes for the pod storage.


This reference architecture creates automatically two different storage accounts for pod storage that will be used in different storage classes to demonstrate the process.

There are two Microsoft Azure storage accounts created as part of the installation process using the ARM template:

  • sapv<resourcegroup> - For the generic storage class (using premium storage)
  • sapvlm<resourcegroup> - To store metrics and logging volumes (using premium storage)

To create more storage accounts, the azure-cli can be used as:

$ azure storage account create --sku-name <sku> --kind "Storage" -g <resourcegroup> -l <region> <storage account name>

This example shows how to create a sapv3sysdeseng storage class using standard storage in the westus region:

$ azure storage account create --sku-name "LRS" --kind "Storage" -g sysdeseng sapv3sysdeseng -l "westus"
info:    Executing command storage account create
+ Checking availability of the storage account name
+ Creating storage account
info:    storage account create command OK

Once the storage account has been created, a StorageClass OpenShift Container Platform object can be created to map it as:

kind: StorageClass
  name: mystorageclass
  storageAccount: sapv3sysdeseng

The cluster-admin or storage-admin can then create the StorageClass object using the yaml file.

$ oc create -f my-storage-class.yaml

Multiple StorageClassess objects can be defined depending on the storage needs of the pods within OpenShift Container Platform.

5.3.2. Creating and using a Persistent Volumes Claim

The example below shows a dynamically provisioned volume being requested from the StorageClass named mystorageclass.

$ vi db-claim.yaml
apiVersion: v1
kind: PersistentVolumeClaim
 name: db
 annotations: mystorageclass
  - ReadWriteOnce
     storage: 10Gi

$ oc create -f db-claim.yaml
persistentvolumeclaim "db" created
$ oc get pvc db
NAME      STATUS    VOLUME                                     CAPACITY   ACCESSMODES   AGE
db        Bound     pvc-be63668e-451e-11e7-b30b-000d3a36dea3   10Gi       RWO           1m

The cluster-admin role can also view more information about the persistent volume

$ oc describe pv pvc-be63668e-451e-11e7-b30b-000d3a36dea3
Name:		pvc-be63668e-451e-11e7-b30b-000d3a36dea3
Labels:		<none>
StorageClass:	mystorageclass
Status:		Bound
Claim:		testdev/db
Reclaim Policy:	Delete
Access Modes:	RWO
Capacity:	10Gi
    Type:		AzureDisk (an Azure Data Disk mount on the host and bind mount to the pod)
    DiskName:		kubernetes-dynamic-pvc-be63668e-451e-11e7-b30b-000d3a36dea3.vhd
    FSType:		ext4
    CachingMode:	None
    ReadOnly:		false
No events.

5.3.3. Deleting a PVC (Optional)

There may become a point in which a pvc is no longer necessary for a project. The following can be done to remove the pvc.

$ oc delete pvc db
persistentvolumeclaim "db" deleted
$ oc get pvc db
No resources found.
Error from server: persistentvolumeclaims "db" not found

Microsoft Azure does not support the Recycle reclaim policy, so all the data will be erased