Frequently asked questions (FAQs) for using Variable Length Deduplication in NetBackup 8.1.1 and later

Article: 100042164
Last Published: 2018-12-14
Ratings: 0 0
Product(s): NetBackup & Alta Data Protection

Description

Starting from NetBackup 8.1.1, you can configure Variable Length Deduplication (VLD) for your backups. VLD can give you better results if your data is not efficiently deduplicated using the existing Fixed Length Deduplication (FLD).

For more information about Variable Length Deduplication, refer to the About variable-length deduplication on NetBackup clients topic from the Veritas NetBackup Deduplication Guide.

Frequently Asked Questions (FAQs)

  • When should I enable VLD?
    If your data is not efficiently deduplicated using the existing FLD, you can enable and configure VLD.

  • Does VLD always give a better deduplication ratio than FLD?
    No. The deduplication ratio of VLD depends a lot on the workload, and the type and content of the data that you want to backup. VLD does not always guarantee a better deduplication ratio than FLD.

  • How does VLD relate to stream handlers and FLD?
    Stream handler technology can take some known workloads VMware or NDMP backups and optimize the data stream to work better than with basic FLD.

  • is VLD better than stream handlers with FLD?
    Testing has shown that when there is a stream handler, performance with FLD is optimal. VLD is best for workloads that do not have a stream handler and also get poor results with FLD.

  • What is important to know before I enable VLD?
    VLD is a processor-intensive deduplication technology. You might get a busy CPU when you back up data using VLD. However, that does not mean that the total throughput decreases. An improved deduplication ratio saves you a lot of disk input output time. In such a case, the total throughput gets better. The overall performance depends on how effective VLD is for you.

  • How can I verify that VLD was used during a backup job?

    1. From the NetBackup Administration Console, double-click the backup job.

    2. In the Job Details dialog box, click the Detailed Status tab and in the Status section, look for the keyword Report=PDDO Stats.
      If you have enabled VLD, VLD Enabled is displayed after the Report=PDDO Stats keyword.

  • What should I do if after I enable VLD, I don't get a better deduplication ratio for the first backup?
    VLD comes largely into play during the second backup because the first backup acts as the baseline. You might not get the desirable deduplication ratio for the first backup.

  • If I don't get the desired deduplication ratio, can I disable VLD?
    You can disable VLD any time and the deduplication reverts to the default FLD mode. It is important that you expire the VLD images to free disk space because FLD backup cannot use the VLD images for deduplicating the data. This scenario might result in a low deduplication ratio.

  • Can I configure VLD on a source MSDP server and duplicate that data to a target MSDP server that uses FLD?
    Yes. In this scenario, VLD is used only in the backup operation and not in the duplicate operation. The duplicate operation just copies the segmented SO from the source MSDP server to the target MSDP server. MSDP does not segment the data during the duplication operation.

  • Can I configure VLD on a source MSDP server and duplicate that data to a CloudCatalyst server (that might then put the data in Access, AWS, etc.)?
    Yes.

  • How can I enable VLD in a load balancing media server scenario?
    NetBackup randomly selects a media server from the load balancing pool for every job. VLD functionality is inside a media server and NetBackup is unlikely to select the same media server everytime. Therefore, when you want to enable VLD in a load balancing scenario, ensure that all the media servers have the same VLD configuration.

  • Can I use VLD with Accelerator?
    Yes. VLD does work with Accelerator. However - it might not be valuable to use VLD with good Accelerator optimization as the extra work would be unnecessary. Most workloads with Accelerator support already get optimal performance with stream handlers and FLD. If Accelerator is enabled with a data set and is not getting good results with a stream handler and FLD, then VLD should be considered.

  • Does VLD support client-side deduplication?
    Yes.

Was this content helpful?