Sign in
Get Help

Deployment Best Practices for Unitrends Backup


Key considerations for successful and optimal deployments of Unitrends Backup virtual appliances, as well as quick tips for product trials.


Key considerations for successful and optimal deployments of Unitrends Backup virtual appliances, as well as quick tips for product trials.

Unitrends Backup may be deployed as a Virtual Appliance in the following environments:

  • VMware vSphere 
  • Microsoft Hyper-V
  • Citrix XenServer
  • Nutanix AHV
  • Amazon AWS
  • Microsoft Azure

*See the Unitrends Compatibility and Interoperability Matrix for specific hypervisor version and feature supportability.

For more information on deployment details, reference the deployment guide specific to your hypervisor, available on our documentation page.

Most trials will leverage the virtual appliances, due to their ease of deployment.

The topics covered include the following:




Basic Steps to Deploy

Virtual appliances are very simple to get up and running:

  1. Download the virtual appliance.
  2. Deploy the virtual appliance and access the web interface via a browser (http://<ip address>).
  3. Configure backup storage under the Configure menu.
  4. Add your protected assets under the Configure menu.
  5. Set up backup jobs under the Protect menu.
  6. (Optional) Register on our forum via the Community dashboard tile.


Choosing Where to Deploy

Unitrends Backup can protect multiple hypervisors and physical servers from a single virtual appliance and scale up to hundreds of TBs of protected data in a single instance. Where you deploy is sometimes dependent on the infrastructure available to you.

Rule of thumb:  Deploy a UB into each virtual environment you are protecting at the hypervisor level.

Following this rule of thumb will typically ensure we are accessing data using the fastest transport modes available for the VMs. Additionally, physical or virtual machines can be protected with agents using any deployment, which simplifies the decision based on the type of hypervisor you use.

Where to Deploy when Protecting MULTIPLE Hypervisors

If you have multiple hypervisors and wish to deploy a UB virtual appliance into each one, you can still centrally manage multiple virtual appliances easily in a single user interface.

When protecting multiple hypervisors in a single instance of Unitrends Backup, it is common to deploy in one hypervisor to protect another. While this does not follow the rule of thumb, it does provide the ability to use one backup instance to protect and deduplicate all your backup data.

In those situations, certain deployments are better than others.  The table below is intended to assist with selecting the most optimal hypervisor when multiple hypervisors are available.

NOTE:  Nutanix AHV virtual machines can be efficiently protected when deployed in other environments.  If you wish to take advantage of the Nutanix storage for backups, you should deploy into the Nutanix AHV cluster.  Best practices should be followed to copy those backups to another location.

Ranking UB Location Comments
Best vSphere Essentials Plus or above
  • Can take advantage of hot-add vSphere backups
  • Good performance for Linux virtual appliances
  • Supports >2TB virtual disks
  • Can protect other assets over the network
  • Can use resources on it's own host for recovery options and CDM testing
Better Free or Essentials vSphere / standalone ESXi
  • Good performance for Linux virtual appliances
  • Can protect assets over the network
  • Free
  • VM restores must go to a licensed host
Good Hyper-V Core 2012 R2 or Higher
  • Supports >2TB virtual disks
  • Can protect other assets over the network
  • Free

Hyper-V 2012
Hyper-V 2008 R2

  • Use only with smaller environments protecting less than 5TB
  • XenServer deployments are required to be used to protect XenServer VMs only.  For protecting other hypervisors or for agent-based protection deployment to other platforms will provide superior results.  
  • Nutanix is a great host platform but lacks IR and replica capability and may cost more in hardware than a standalone host to run a UB.  
  • AWS/Azure have cost, performance, and scalability limitations vs other platforms but are required to protect systems inside those environments.  Customers may find Unitrends Backup for Azure superior to a UB in Azure.  



Configuring Backup Storage

Storage Protocols

Unitrends Enterprise Backup can perform advanced, adaptive inline deduplication. Therefore, expensive deduplication storage devices are not required to obtain deduplication ratios of 20-to-1 or more. However, it is important that your storage is capable of processing the IO required for global deduplication across all backups.  For this reason, we do not support the use of 3rd party native deduplication hardware for UB backup storage.  (Such devices are fully supported for Cold Copy archive data).  

As a result, we recommend high-speed drives and highly reliable network connectivity to your backup storage.  Unitrends Onboarding will expect the use of host-managed datastores and virtual disks as a best practice, any other use should be communicated to onboarding in advance.  

The good thing is that pretty much all storage vendors can be configured to leverage Unitrends deduplication effectively. This is really a matter of using the best practice protocol. Below is a list of rankings for backup storage protocols to use when configuring your backup storage for optimal performance, footprint, and scalability.

Ranking Storage Protocol Comments
Best (by far) Attached virtual disk
  • Uses the hypervisor IO stack for interaction with backup storage
  • Large virtual disks are available
  • Easily expanded by adding virtual disks
  • Easily migrate an existing appliance to new hosts or new storage
  • Nearly all NAS and SAN storage vendors can be configured as virtual storage through NFS, iSCSI, Fiber Channel and even SMBv3 (SMB3 not recommended)
  • Types of storage protocols connecting to the hypervisor can also be ranked as follows:
    • (Best) Direct attached storage
    • (Better) Fibre Channel and iSCSI
    • (Good) NFS VMWare datastores
    • (Not Rec.) SMB3 storage
Good NFS within the UB guest
Okay iSCSI within the UB guest
CIFS (non-Windows)
  • iSCSI has good performance but lacks flexibility and expansion.
  • Direct iSCSI will not perform as well as hypervisor connected iSCSI Datastores
  • CIFS is supported, but typically slower performance and less reliability than other protocols and has security risk if not network segregated from other windows machines. 
  • Typically used only for smaller data sets
  • For a list of supported vendors in this configuration, see Supported external storage vendors for use with Unitrends Enterprise Backup appliances
Do Not Use CIFS on Windows  or 
More that one storage device on a single UB
  • Using an NTFS Fat32 file system hosted by a Windows OS has been confirmed to result in data loss and is not supported for backup or cold copy data.
  • Consult a Unitrends representative, but it is much better to convert the Windows server to a Hyper-V host or an ESXi host to run the UB virtual appliance
  • Use of multiple storage devices, such that one may lose power while another is online, will result in data loss of deduplicated data systems. 


Deployment Tip:
Deduplication performance is most limited by IO performance.  The Unitrends Database is deployed by default to the configured backups device.  If that device has slow performance, the database can be moved to higher performance virtual disk attached to the UB.  See the deployment guide or discuss this option with your professional onboarding engineer for more info.  


Third Party Deduplication Devices

Some storage vendors offer “smart” storage that performs its own deduplication efforts on data written to it. Unitrends supports these devices for Cold-Copy archive storage, but does not support them for primary backup storage. 


Backup Mode Selection

Unitrends supports a number of backup modes to ensure flexible protection policies for various types of data. A single job can use one backup mode, but UB can leverage multiple backup modes across various jobs. The below table will help you choose the best mode for your strategy.


Ranking Backup Mode Comments
Best Incremental Forever
  • Provides the fastest backup window after the first full backup
  • Recommended for VMware, Hyper-V, Nutanix AHV and most file-level backups
  • Reads the full disk once and then processes only changes going forward
Better Full / Incremental
Full / Transaction
  • Recommended for Exchange backups
  • Recommended if you want to control when full backups are taken for the purpose of backup copy management
  • Recommended when you want to force a full read of all data periodically
  • Inline deduplication ensures that even full backups only write changes to the backup storage
  • Full + Transaction is recommended for SQL Databases with the Full recovery model enabled.  
Good Full / Differential
  • Recommended for SQL backups with databases in simple recovery model. (full recovery model will require full + transaction backup types)
  • Recommended if you want to simplify recovery of backup copies from tape at the expense of longer backup copy times compared to full / incremental
  • Inline deduplication ensures that even full backups only write changes to the backup storage
Okay Fulls
  • Recommended for Citrix XenServer backups
  • Recommended when RPOs are very long (one week or longer)
  • Can be used with Incremental Forever if you only want full backups to be periodically copied to backup copy storage
  • Inline deduplication ensures that even full backups only write changes to the backup storage


Trial Tip:
To be sure you see the full benefits of Unitrends best-in-class deduplication:
  1. Run multiple jobs with multiple machines in each job.
  2. Be sure to retain a number of backup sets. The more backups stored, the better the deduplication ratio.


Sizing Your System

CPU, memory and backup storage configurations are critical for proper operation and performance. With UB virtual appliances, these can very easily be increased due to the benefits of virtualization. This makes the process of tuning your backup configuration much simpler than any other deployment type.

By default, Unitrends Backup deploys with 2 vCPUs and 8 GB of RAM which is sufficient for basic deployment processes, but is not intended to be a run-time configuration for backups (see subsection below titled "CPU and Memory").  It is expected you will need to modify these defaults based on your protection requirements.  A minimum of 200 GB of backup storage is required for operation. 

A general rule of thumb is to configure a UB with similar specs to what would be included in an equivalent Unitrends Recovery Series appliance, though more specific details are described below.  

Backup Storage 

Rule of thumb:  Assuming the use of Advanced Adaptive Deduplication (Deduplication Level 3 on virtual appliances) and incremental forever scheduling, use 1.5 times the protected capacity for your backup storage size as a minimum.  This rule can vary depending on how well the data can deduplicate, compression capability of your data set, change rates of the data set, or if you have retention requirements that are beyond 30 days.  For Deduplication Level 1 or 2, substantially more storage may be required.  
Adding more storage: If you deployed using Attached virtual disk option, you can increase the storage pool for backups by adding another virtual disk to the UB VM. Afterward, log into the Unitrends HTML5 /ui/ Web Admin Interface and verify there are no running Jobs. Then click on CONFIGURE > Storage, and select the name Internal and click the Edit button. Select the newly added disk and click Add, then save.

Note:  Unitrends Forever Cloud is a great way to offload long-term retention to Unitrends Cloud services and free up resources at your local site.

CPU and Memory

Because Unitrends offers our enterprise software pre-packaged with our Recovery Series physical appliances, you can see how those physical appliances are configured in terms of CPU, memory, and backup storage capacity using the Recovery Series Models table.  

When using all relevant enterprise features, such as inline deduplication, forever incremental, backup copies and encryption, we recommend a typical production configuration to use 4 vCPUs and 16 GB at a minimum. The more memory the better, but no less than 4GB of RAM per allocated vCPU.  

Smaller environments may be able to use fewer resources.

For most environments, the rule of thumb is 1 vCPU and 4 GB of RAM per active concurrent task (backup, backup copy, replication thread, or restore).  By default, UB is configured for 10 concurrent backup tasks max, and this can be lowered to limit resource requirements.  This can also be increased by our support team if necessary beyond 10 threads.  However, that's only backup threads.  Hot copies, cold copies, FLR processes, and restores also require threads.  With release 10.0 we also adding Automated Provisioning and testing services to Enterprise Plus licensed units and if that service is in use additional CPUs and RAM may be necessary per concurrent restoring asset. 
Trial Tip: 
Check the built-in Appliance Load Report under the Reports menu of the user interface to see if your system requires additional resources. You may also find it useful to leverage your hypervisor monitoring tool to determine if the UEB virtual appliance is getting the resources it needs to meet your backup windows.



How to Scale to Meet Your Needs

Unitrends Backup can scale up significantly. It can also scale out, while still being managed by a centralized user interface. In fact, you can currently protect at least 6 petabytes (PB) of data in a single, optimally configured deployment by using multiple, large UB virtual appliances.

As you protect more data, 10 Gb Ethernet or 8 Gb Fibre Channel connections to your backup storage system become much more important to ensure your backup windows are being met. Fast, enterprise disk drives may be required for the use of our database to achieve the best performance.  Note, all of your storage does not need to be fast, only the portion that contains the Unitrends Database.  It is easy to move the Unitrends Database to faster storage following this KB: Move the Unitrends database off of stateless storage and onto a new partition

When scaling up your backup storage within a single UB virtual appliance, you can use very large virtual disks in your initial deployment or simply add more virtual disks. When adding virtual disks, UB will pool them together to create a single, large backup storage pool.

If you choose to add virtual disks for scaling, here are a few guidelines for optimal reliability and performance:

  1. It is highly recommended to add virtual disks from physical storage that resides in the same chassis to avoid multiple points of failure for your backups.  Virtual disks do not need to be in the same LUN, but, if some storage is available and other is offline, this is highly likely to cause data loss and it never a recommended configuration.  When using deduplication the use of multiple separate physical storage devices may be denied support instead requiring the deployment of individual UB appliances connected to each device independently.  See Risk below.
  2. Ensure all virtual disks for backup storage use underlying storage with the same performance class to avoid undesirable performance impacts.  If disparate storage must be used, it is recommended to deploy additional UB appliances as opposed to mixing storage in a single unit.  Remember there is no charge for deploying additional units.  
  3. Use Thick Provisioned or Lazy Zero disks.  if a UB is over-provisioned and a LUN/datastore runs out of space the system may crash resulting in data loss.  Thin provisioned disks are not recommended. 

If you choose to use pass-thru storage over NFS, iSCSI, or CIFS connected directly to your UB instance, the following guidelines apply

  1. Do not use a UB's backup share for any other purposes, the share must be dedicated to that UB and only that UB.  Any non-Unitrends data in that share will be detected and removed without warning.  
  2. Multiple shares in the same NAS device should only be leveraged if each occupies an independent NAS volume or in cases where each share reports free space independently of other shares based on specified share size. 
  3. No other systems or UBs should leverage other shares in the same volume for storage.  Reduction of the free space available to a UB by outside influences may result in unplanned purging or data, backup failure, or complete UB data loss.  Each deployed UB requires it's own dedicated storage volume or share.
  4. Use of multiple separate NAS devices for backup storage on the same UB is not supported where deduplication is in use
  5. increasing the size of your NAS through vendor tools by growing the volume is supported, do so only when the UB is powered off.  iSCSI LUNs can be increased in size but additional space will only be seen after additional manual steps.  


 Risk: If one virtual disk fails in the backup storage pool, all backup data is impacted.
Mitigation: Leverage Unitrends Backup Copy jobs to ensure that you have multiple copies of your backup data on different media and different locations, such as the Unitrends Cloud. This type of strategy becomes critically important when you add multiple virtual disks to expand your backup storage.


Have more questions?

Contact us

Was this article helpful?
0 out of 0 found this helpful

Provide feedback for the Documentation team!

Browse this section