Deduplication Appliance Best Practices

Backup Job Settings

Advanced Settings

Below are the general recommendations for the Advanced Settings (Storage > Advanced):

Backup Tab

In general, deduplication devices have the lowest performance during read operations. As such, to maximize performance, the backup jobs should be write-only. Configure the backup job to use (Forward) Incremental with weekly Active full backups to achieve this.

Note: Some Dedupe Appliances that integrate with Veeam Backup & Replication can perform Synthetic Full operations.

• Backup Mode
  • (Forward) Incremental
    
• Active full backup
  • Enabled and set to weekly.
    Weekly Full restore points will ensure that during a restore, as few restore points must be read from as possible.

Maintenance Tab

• Perform backup files health check
  • Disabled
    While enabling this feature with dedupe storage is possible, it requires many read operations, which will take a very long time to complete. With larger backup files on deduplicated storage, this feature could cause a backup job to spend days performing the health check, causing the job to not create new restore points.
• Defragment and compact full backup file
  • Disabled
    There is no need for this option when Active Fulls are being created.

Storage Tab

• Inline data deduplication
  • Disabled
    Disabling deduplication in Veeam Backup & Replication will cause the restore points to appear larger, but because Veeam does not deduplicate them, it will allow the storage's deduplication function to be more effective.
    
    Note: Storage that utilizes a non-dedupe zone for storing initially written files, such as ExaGrid's Landing Zone, may benefit from having Veeam's deduplication enabled to allow more restore points to be held in that rapid access state before being processed by the storage deduplication system.
• Compression level
  • Optimal or Dedupe-friendly
    The Repository settings section of this article advises enabling a function that will cause all restore points to be decompressed before being written. As such, the compression set at the job level will only impact data compression while backup data is being transferred across the network. To reduce network congestion, either Optimal or Dedupe-friendly compression may be used. The latter will slightly reduce the effort needed to decompress before writing. Compression at the job level should rarely be completely disabled as it will significantly increase the amount of data that must be transferred over the network to the machine writing the data to the backup repository.
• Storage optimization (Block Size)
  • 4MB
    This block size, formerly called "Local target (large blocks)," will help improve performance when storing backup files on deduplication storage as it reduces the size of the backup file's internal metadata table.
    
    Note: For storage that utilizes a non-dedupe zone for storing initially written files, such as ExaGrid's Landing Zone, it is beneficial to set the Block Size to 1MB. A 1MB Block Size will cause Inline Data Deduplication to be more effective, thereby allowing more restore points to be held in that rapid access state before being processed by the storage deduplication system.
• Encryption
  • Disabled
    Data encryption has a negative effect on the deduplication ratio if you use a deduplicating storage appliance as a target. Veeam Backup & Replication uses different encryption keys for every job session. For this reason, encrypted data blocks sent to the deduplicating storage appliances appear different, even though they may contain duplicate data. If you want to achieve a higher deduplication ratio, you can disable data encryption. If you still want to use encryption, you can enable the encryption feature on the deduplicating storage appliance itself.

Backup Copy Job Settings

Backup Chain Settings

When targeting deduplicated storage, the backup copy job should be configured as follows to force the backup copy job to perform primarily write-only operations:

• Keep certain fulls backups longer for archival purposes
  • Enabled and set to have at least 1 weekly
    Enabling this option forces the backup copy job to enforce retention using the forward-incremental retention policy, which prevents the job from using the non-GFS retention method that involves merging the oldest increment into the full (a retention method known as Forever Forward Incremental). Forever Forward Incremental retention method is suboptimal for deduplication storage as it involves many small read and write operations to enforce retention.
• Read the entire restore point from source backup instead of synthesizing it from increments
  • Enabled*
    Enabling this option for local-to-local backup copy jobs being written to deduplication storage that does not support block cloning will force the job to perform strictly write-only operations to create the GFS full. Thereby avoiding unnecessarily taxing synthetic GFS full creation processing that involves creating the GFS full restore point by reading from previous restore points stored on the dedupe storage.
    
    *This option should be disabled if:
    • The backup copy job will be transferring backups across a low-speed connection. Specifically, a link so slow that transferring the entire full to the destination would be slower than the job creating a full synthetically using data already present on the dedupe storage.
      _{Some offsite connections may be fast enough to perform the full transfer faster than the synthetic creation process. It will come down to comparing the network throughput vs. the read+write speed of the dedupe storage. For example, if the connection used to reach the backup copy destination is 100Mbps (12.5MB/s), and the dedupe storage can create a synthetic full at 40MB/s, the option should be disabled. However, if the connection is 500Mbps (62.5MB/s), that is faster than the dedupe storage in this example, so reading the entire restore point across the network would be quicker.}
    • The target deduplication appliance backup repository utilizes block cloning functionality.
      Block cloning is available with integrated deduplication appliances (HPE StoreOnce Catalyst, Exagrid, Data Domain DDboost, and Quantum DXi)

Advanced Settings

Below are the general recommendations for the Advanced Settings (Target > Advanced):

Maintenance Tab

• Perform backup files health check
  • Disabled
    While enabling this feature with dedupe storage is possible, it requires many read operations, which will take a very long time to complete. With larger backup files on deduplicated storage, this feature could cause a backup job to spend days performing the health check, causing the job to not create new restore points.
• Defragment and compact full backup file
  • Disabled
    There is no need for this option when Active Fulls are being created.

Storage Tab

• Inline data deduplication
  • Disabled
    Disabling deduplication in Veeam Backup & Replication will cause the restore points to appear larger, but because Veeam does not deduplicate them, it will allow the storage's deduplication function to be more effective.
    
    Note: Storage that utilizes a non-dedupe zone for storing initially written files, such as ExaGrid's Landing Zone, may benefit from having Veeam's deduplication enabled to allow more restore points to be held in that rapid access state before being processed by the storage deduplication system.
• Compression level
  • Auto
    The Repository settings section of this article advises enabling a function that will cause all restore points to be decompressed before being written. As such, the compression set at the job level will only impact data compression while backup data is being transferred across the network.
• Encryption
  • Disabled
    Data encryption has a negative effect on the deduplication ratio if you use a deduplicating storage appliance as a target. Veeam Backup & Replication uses different encryption keys for every job session. For this reason, encrypted data blocks sent to the deduplicating storage appliances appear as different though they may contain duplicate data. If you want to achieve a higher deduplication ratio, you can disable data encryption. If you still want to use encryption, you can enable the encryption feature on the deduplicating storage appliance itself.