Showing results for 
Search instead for 
Did you mean: 

NetBackup 10.4 Lets K8s Backups Fly Even Higher

Level 4

NetBackup 10.4 Lets K8s Backups Fly Even Higher with Accelerator Backups and Improved Namespace Component Restore Ordering --- Turbo-Supercharge Your K8s Protection Jobs


K8s data sets are exploding in size and our users needed some serious throughput tools to keep up. We’re seeing 10 TB+ backups for K8s at user sites with no slowdown in growth.  NetBackup 10.4 includes great new features to really help turbo-supercharge the data out of K8s in record time, then put it back again in the startup/operations order you specify. Deduplication, incremental backups, Accelerator, and advanced namespace restoration support now all work together to provide the ultimate K8s data protection optimization. Even the largest modern K8s data protection jobs can be handled quickly with minimal resource utilization and restored in the proper order to come back online quickly.

Accelerator Support for K8s Data

Version 10.4 Accelerator-based backups function similarly to how NetBackup does cloud workloads that also use our elastic data mover (scales on demand) framework.  We send only changed information across the network but still catalog the backup image as a full backup.

Also, any deduplicated media server can perform Accelerator backups without having to start a new change tracking log, which triggers a full backup instead of an incremental.  Selecting a snapshot from backup and deduplicated target storage activates the use of Accelerator as shown in the figures below:



Note that as in previous NetBackup versions, Accelerator requires storage space for a change tracking log.  If K8s log storage has not been provisioned for NetBackup, then NetBackup 10.4 defaults to a traditional full (every data block copied and moved) backup.

Network Restoration Optimizations for K8s Restore Data

Users said they wanted to control the restoration order of K8s elements. Modern K8s configurations are very complex, especially the networking configurations. Pods need network definitions to be in place first before they’re restored and operational.  These factors mean restores can fail if namespace components aren’t brought up in the correct order.

To prevent restores from failing this way, NetBackup 10.4+ allows you to choose to have network definitions restored first. It does this with the Kubernetes restore order list as shown in the figures below. Previous NetBackup versions allowed users to select which K8s resource types were included in a restore only. Version 10.4+ now allows you to configure not only the resource types, but the order in which the resource types are restored as shown as in the following two figures. Note how the restore order can be edited in case it’s found that previous restore attempt configuration(s) were unsuccessful for K8s technical reasons:



Confirmation of this restore order is echoed in the restore job details as shown in this figure:


Of course, K8s requirements for a valid restore, such as restoring to compatible types of destinations and resources still have to be followed. However, assuming that’s the case this new feature will greatly improve the viability and success rate of K8s restores.

How This New Combination Performs

So, how does this turbo-supercharged combination perform? The following charts are from test data taken from K8s Accelerator-enabled jobs. These are lab-derived figures to show the general performance trend you should expect. However, the specific time savings you see will vary by K8s configuration test and resources available at the time.

The first chart shows baseline snapshot times in blue and the actual backup file system (BFS) times in orange. Full and differential incremental (DINC) backups have been performed. The data change rate is only 1.1%. This data is for a K8s backup without Accelerator activated. As should be expected, snapshot times vary very little. De-duplication reduces the time for the second full backup:


The second chart shows another round of jobs with Accelerator activated and the same data change rate. Note how not only the incremental backups, but the second full backup are much quicker than the initial full backup.


The third chart shows another round of jobs with Accelerator activated and a much higher 6.5% data change rate. The second full is now almost the same size as an incremental with de-duplicated target storage working together with Accelerator.


So, upgrade to NetBackup 10.4 today to turbo-supercharge your K8s backups. Save storage, network, and time resources all while protecting K8s faster and better than ever before.