Upgrading to Unitrends Virtual Backup v8
Part of my responsibility is supporting our various Backup systems and for handling VM backups in the VMware world, there's 2 clear contenders winning the popular vote; The most popular is Veamm Backup&Restore, and the other popular option is Unitrends Virtual Backup (Formerly PHD Virtual Backup). I want to talk about my experience migrating from PHD v6.5 to UVB v8.
UVB uses virtual appliances that you deploy, much like earlier versions of PHD and UVB. If it's your first appliance, all you have to do is specify an IP Address, go to that address in a web browser, and follow a short setup guide. Once the first appliance has been deployed, all others are configured through the web interface on the first. Unlike previous versions of PHD or UVB, each appliance you deploy plays a role. Appliances can hold Presentation, Management, or Engine roles. A Presentation appliance simply hosts the web inter face through which all other configuration and management is performed. You only need one of these. A Management appliance just acts as the primary appliance in the UVB environment. Again, you only need one of these per UVB environment. When I say UVB environment specifically, rather than just referring to your environment is because in UVB v8, VMWare, Citrix XenServer and HyperV backups are all supported. For example, if you have VMWare and HyperV in your business that you have to backup, you would only need 1 Presentation appliance, and 2 management appliances; 1 to be the primary appliance for the VMWare environment and a separate manager to be the primary appliance for the HyperV environment. UVB v8 refers to each as environments even though you may very well have multiple hypervisors in your actual IT environment.
For storage, you can use NFS mounts if you wish but Unitrends will give you a handy-dandy 30 page document explaining that block-level storage is faster than file-level storage, in case you didn't already know that. We created LUNs, assigned the LUNs to datastores & attached virtual disks to the VBAs using those datastores. (There's probably a better way to word that but I'm still getting my feet wet in VMware.)
There's 4 different types of storage you can configure on your backup appliances; Backup, Archive, Sync & Instant Recovery Write Space. Backup Storage is exactly what it sounds like. Nothing particularly special about it. Archive Storage is essentially a backup of your backups. Unlike your standard Backup Data Stores (BDS), Archive Storage will usually have longer retention range than Backup Storage. Sync storage makes a copy of your data to a separate BDS. Sync storage writes changes from recent backup to a VM back to a replica of your BDS, including retention settings of the originating BDS. The fourth & likely most interesting storage type you can add to your engine appliance is Instant Recovery (IR) Write Space. Normally, when you restore or recover a VM, the appliance will build a VM from it's replica + whatever snapshot you chose and place it on the same host & datastore where the originating VM was backed up from. Writing that VM back to the original datastore can take some time and time can be crucial when you're restoring a VM of a critical service. It's a temp space available for you to get a VM (or several) back up & running. Once a VM has been IR'd, you could simply s-vMotion it to a production datastore but UVB Motion Seeding is a more automatic way to do this. It work in two stages behind the scenes. First is the seeding. It's basically a standard restore that is initiated during an IR job. Second is the UVB Motion. This syncs all changes made on the temp space to the restored VM on the datastore you specified during the UVB Motion Seeding portion of the IR. IR Write Space can be added to any appliance with the Engine role at anytime but the other Engine appliances can have Backup, Archive OR Sync storage. (I wanted to spell that out in case the Admin guide didn't make that crystal clear.)
We had some questions when we upgraded. For example, in the older version of PHD, you specified jobs for a VBA (VBA1 handles Backup Job 1, VBA2 handles Backup Job 2, etc.) but that's changed in UVB v8. All available VBAs tackle any job that starts. Another example was setting VMs to backup with CBT (Change Block Tracking). In the old version we had to specify each each VM be backed up with CBT but in the latest version, all powered-on VMs are backed up with CBT enabled by default. (With that said, powered off VMs will be backed up in full.)
Other than some of the changes & new features described above, UVB v8 is actually quite nice. The web interface is very intuitive, it's stable and has some awesome reporting features. If you're upgrading to v8 or trying UVB for the first time, I hope you found this information helpful.
UVB uses virtual appliances that you deploy, much like earlier versions of PHD and UVB. If it's your first appliance, all you have to do is specify an IP Address, go to that address in a web browser, and follow a short setup guide. Once the first appliance has been deployed, all others are configured through the web interface on the first. Unlike previous versions of PHD or UVB, each appliance you deploy plays a role. Appliances can hold Presentation, Management, or Engine roles. A Presentation appliance simply hosts the web inter face through which all other configuration and management is performed. You only need one of these. A Management appliance just acts as the primary appliance in the UVB environment. Again, you only need one of these per UVB environment. When I say UVB environment specifically, rather than just referring to your environment is because in UVB v8, VMWare, Citrix XenServer and HyperV backups are all supported. For example, if you have VMWare and HyperV in your business that you have to backup, you would only need 1 Presentation appliance, and 2 management appliances; 1 to be the primary appliance for the VMWare environment and a separate manager to be the primary appliance for the HyperV environment. UVB v8 refers to each as environments even though you may very well have multiple hypervisors in your actual IT environment.
For storage, you can use NFS mounts if you wish but Unitrends will give you a handy-dandy 30 page document explaining that block-level storage is faster than file-level storage, in case you didn't already know that. We created LUNs, assigned the LUNs to datastores & attached virtual disks to the VBAs using those datastores. (There's probably a better way to word that but I'm still getting my feet wet in VMware.)
There's 4 different types of storage you can configure on your backup appliances; Backup, Archive, Sync & Instant Recovery Write Space. Backup Storage is exactly what it sounds like. Nothing particularly special about it. Archive Storage is essentially a backup of your backups. Unlike your standard Backup Data Stores (BDS), Archive Storage will usually have longer retention range than Backup Storage. Sync storage makes a copy of your data to a separate BDS. Sync storage writes changes from recent backup to a VM back to a replica of your BDS, including retention settings of the originating BDS. The fourth & likely most interesting storage type you can add to your engine appliance is Instant Recovery (IR) Write Space. Normally, when you restore or recover a VM, the appliance will build a VM from it's replica + whatever snapshot you chose and place it on the same host & datastore where the originating VM was backed up from. Writing that VM back to the original datastore can take some time and time can be crucial when you're restoring a VM of a critical service. It's a temp space available for you to get a VM (or several) back up & running. Once a VM has been IR'd, you could simply s-vMotion it to a production datastore but UVB Motion Seeding is a more automatic way to do this. It work in two stages behind the scenes. First is the seeding. It's basically a standard restore that is initiated during an IR job. Second is the UVB Motion. This syncs all changes made on the temp space to the restored VM on the datastore you specified during the UVB Motion Seeding portion of the IR. IR Write Space can be added to any appliance with the Engine role at anytime but the other Engine appliances can have Backup, Archive OR Sync storage. (I wanted to spell that out in case the Admin guide didn't make that crystal clear.)
We had some questions when we upgraded. For example, in the older version of PHD, you specified jobs for a VBA (VBA1 handles Backup Job 1, VBA2 handles Backup Job 2, etc.) but that's changed in UVB v8. All available VBAs tackle any job that starts. Another example was setting VMs to backup with CBT (Change Block Tracking). In the old version we had to specify each each VM be backed up with CBT but in the latest version, all powered-on VMs are backed up with CBT enabled by default. (With that said, powered off VMs will be backed up in full.)
Other than some of the changes & new features described above, UVB v8 is actually quite nice. The web interface is very intuitive, it's stable and has some awesome reporting features. If you're upgrading to v8 or trying UVB for the first time, I hope you found this information helpful.
Comments
Post a Comment