Network-optimized data transfer improves backup and restore performance for Kubernetes workloads that use PersistentVolumeClaims (PVCs). Instead of routing data through the Kubernetes API server, data moves directly between the cluster and backup infrastructure.
This approach is recommended for large or stateful workloads because it:
-
Reduces load on control plane components
-
Improves backup and restore speed
-
Lowers data transfer costs in cloud environments
Important
-
For new clusters, network-optimized backups and restores are enabled by default.
-
For existing clusters, after enabling the feature, you must run full backup of the cluster.
How network-optimized data transfer works
Commvault uses temporary pods (cv pods) in the Kubernetes cluster to move data.
-
During backup, the temporary pods read data directly from PVCs. Data is sent to the MediaAgent.
-
During restore, the temporary pods retrieve data from the MediaAgent. Data is written directly to PVCs.
Recommended use cases
Use network-optimized data transfer in the following scenarios:
-
Clusters with large PVCs or stateful applications
-
Environments where API server throughput is limited
-
Cloud environments where direct data transfer reduces cost
-
Restores of large datasets or cross-cluster restores
-
Restricted or air-gapped environments with access to required container images.
Requirements
-
The cv pods must be able to connect to the CommServe server, access node, and MediaAgent,either directly or through a configured network gateway.
Network configuration details
To enable communication between the Kubernetes cluster and backup infrastructure, configure server groups and a network gateway topology.
-
Create a server group for the Kubernetes cluster pseudo client.
-
Create a server group for the network gateway clients that are reachable from the Kubernetes cluster, and can connect to the CommServe server, access node, and MediaAgent.
-
Create a server group for infrastructure clients that initiate connections to the network gateway group. For information about creating a server group, see Adding a Server Group by Using Manual Association.
-
Then configure a network gateway topology using these server groups. For more information, see Setting Up a Network Gateway Connection Using a Predefined Network Topology - Client Type is Servers.
-
-
The required container image must be accessible to the cluster. Commvault uses the cvk8sfs image for temporary data movement pods. You can also pull the image from Docker Hub.
If you have an air-gapped cluster, you can specify a private container registry that contains the image.
-
The Kubernetes cluster must have sufficient memory available for the cv pods that use the File System Agent to back up and restore PVC data. Each cv pod requires the following memory resources:
-
A minimum of 2 GB of RAM for up to 500,000 folders in the backup data.
-
An additional 1.5 GB of RAM for every additional 500,000 folders in the backup data.
For information about configuring resource limits, see Modifying the Resource Limits for Commvault Temporary Pods for Kubernetes.
-
Enable network-optimized data transfer
To enable network-optimized backups and restores for existing clusters, configure the Network Optimized Backups setting on the cluster or application group.
-
From the Command Center navigation pane, go to Protect > Kubernetes.
-
On the Clusters tab, click the cluster.
-
On the Configuration tab, in the Settings section, click Add.
-
Click Lookup.
-
Select the check box next to Network Optimized Backups, and then click Select.
-
Enable the Value toggle key.
-
In the Comment box, enter a description.
-
Verify that the Enable check box is selected, and then click Save.
-
From the Command Center navigation pane, go to Protect > Kubernetes.
-
On the Application groups tab, click the application group.
-
On the Configuration tab, in the Settings section, click Add.
-
Click Lookup.
-
Select the check box next to Network Optimized Backups, and then click Select.
-
Enable the Value toggle key.
-
In the Comment box, enter a description.
-
Verify that the Enable check box is selected, and then click Save.
Considerations
-
Restores automatically use network-optimized data transfer when backups use this feature.
-
Client-side processing increases CPU usage on cluster nodes. You can disable deduplication or compression to reduce CPU usage. For more information, see Modifying the Deduplication Properties for the Client.