Faster Incremental Backups Using the NetApp SnapDiff Feature

You can use SnapDiff for faster incremental backups for regular backups of NAS file servers (network shares). SnapDiff is a NetApp feature that identifies which files changed between snapshots, so that backups do not have to traverse the entire file system to determine which files changed.

On NetApp storage arrays, all IntelliSnap backups of network shares use SnapDiff. You can also use SnapDiff for regular backups of network shares.

SnapDiff is especially effective when you have a large number of files that do not change a lot between snapshots.

NetApp SnapDiff Support

*To run SnapDiff v1 or v2 on this version, contact NetApp support to restore the APIs and contact Commvault support also. If back up jobs were already run on subclients and you upgraded ONTAP to either of these versions, then the first incremental scan will take longer as a TrueUp operation runs to gather file information. All the subsequent jobs will continue and use a recursive scan to collect file information.

**If back up jobs were already run on subclients and you want to run SnapDiff v1 or v2 or v3 for existing subclients after you upgraded ONTAP to either of these versions, then contact Commvault support.

How SnapDiff Works

The first full backup job creates a baseline snapshot. A snapshot is created, data is scanned by traversing the file system, a snapshot is created, and then data is backed up from the snapshot.

Subsequent backups are incremental. SnapDiff identifies the data that changed by comparing the previous and current snapshots, without traversing the file system. Data is then backed up from the snapshot.

At the end of each job, snapshots created by the previous job are deleted from the array, and only one set of snapshots is retained on the array.

The scan operation that traverses the file system is used only in these cases:

• The snapshot from the previous job is not available

• The snapshot creation fails in the current job

• The SnapDiff operation fails

Configuration Requirements for SnapDiff V2

1. Verify RPC is enabled on the vServer being cataloged or backup copied. Even if a cluster client is being cataloged, the vServer for the volume being cataloged or backup copied must have RPC enabled. On the file server, verify RPC is enabled on the vServer using the following commands:

   set advanced

   vserver snapdiff-rpc-server show -vserver <vserver>

   vserver snapdiff-rpc-server on -vserver <vserver>

2. Verify whether an NFS interface exists between the vServer and the Commvault MediaAgent. Specifically, confirm that there exists a vServer NFS interface on the file server node containing the volume being cataloged using the following commands:

   vserver nfs show -vserver <vserver>

   network interface show -vserver <vserver> -data-protocol nfs

Configuration Requirements for SnapDiff V3

1. Verify RPC is enabled on the vServer being cataloged or backup copied. Even if a cluster client is being cataloged, the vServer for the volume being cataloged or backup copied must have RPC enabled. On the file server, verify RPC is enabled on the vServer using the following commands:

   set advanced

   vserver snapdiff-rpc-server show -vserver <vserver>

   vserver snapdiff-rpc-server on -vserver <vserver>

2. Verify whether an NFS interface exists between the vServer and the Commvault MediaAgent. Specifically, confirm that there exists a vServer NFS interface on the file server node containing the volume being cataloged using the following commands:

   vserver nfs show -vserver <vserver>

   network interface show -vserver <vserver> -data-protocol nfs

3. Verify that the cluster array login is authorized for REST, HTTP, and SSL. For more information on setting up a NetApp cluster login.

   1. To enable REST API based authorization to the cluster array, run the following commands:

      vserver services web access show -vserver <vserver> -name rest -role <role>

      vserver services web access create -vserver <vserver > -name rest -role <role>

   2. To enable HTTP and SSH based authorization to the cluster array, run the following commands:

      security login show -vserver <vserver> -user-or-group-name <user> -application http

      security login create -vserver <vserver> -user-or-group-name IntelliSnapUser -application http -authentication-method password -role <role>

      security login show -vserver <vserver> -user-or-group-name <user> -application ssh

      security login create -vserver <vserver> -user-or-group-name IntelliSnapUser -application ssh -authentication-method password -role <role>

   3. To enable the SnapDiff and license APIs for the cluster array, run the following commands:

      security login role show -vserver <vserver> -role <role> -cmddirname "snapdiff"

      security login role create -vserver <vserver> -role <role> -cmddirname "snapdiff" -access all

      security login role show -vserver <vserver> -role <role> -cmddirname "license"

      security login role create -vserver <vserver> -role <role> -cmddirname "license" -access all

4. Note: For SnapDiff v3, the array management entries for the vServer must have tunneling to the selected cluster. SnapDiff V3 requires security key authentication that is supported only on the cluster interface.

Controlling the Behavior of the SnapDiff Feature

If the snapshot creation fails or if the snapshot path is not accessible in the current job, then data is scanned and backed up from the file system instead of from snapshot path. You can use the additional setting bBackupFromSnapshotOnly to back up data from the snapshot path only. If the additional setting is set, then when there are paths where the snapshot creation fails or when the snapshot path is not accessible will be reported as failed.

Related Topics

• Read about the system requirements for SnapDiff.

• To get started with Network Share, see Network Share.

• To learn how to control the behavior of the SnapDiff feature, see controlling the behavior of the SnapDiff feature.

• To compare the performance of backup operations with SnapDiff against backup operations without SnapDiff, see SnapDiff Feature Performance Numbers.