In an earlier blog we explored vRanger server-based backup and restore architecture options, where all backup and restore data handling is performed by the vRanger server. As a quick reminder, there are different data paths available when vRanger is installed in a virtual machine (VM) or on a physical server. When the vRanger server is installed in a virtual machine (VM) it uses the SCSI HotAdd VDDK advanced transport method, and when it’s installed on a physical server it uses the SAN advanced transport method. In both cases it can also use the Network transport method as a failover transport method, or instead of either of the advanced transports.
A new backup architecture option is available starting with vRanger v5.4: Virtual Appliance (VA)-based backup and restore. The vRanger VA can still be used for replication and Linux-based file level restores, but in this blog we’ll focus on the new VA-based backup and restore architecture option. The vRanger VA-based backup and restore configuration is illustrated below in Figure 1.
The vRanger VA is a lightweight virtual machine (VM) which includes a simple deployment of Linux that serves as an operating environment for the components needed to complete backup and restore operations (along with the aforementioned replication operation). With VA-based backups and restores, data is handled by the VA instead of the vRanger server. This distributes job execution and data handling out from the vRanger server to the VA’s, and effectively moves backup data to the backup repository in a “direct to target” fashion.
The vRanger server is still installed either in a VM or on a physical server, and from there VA’s can be centrally deployed through a simple and efficient Wizard to stand alone hosts or, if you have a clustered environment, one or more hosts in the cluster. One VA can be shared by all hosts in the cluster, and additional VAs can optionally be deployed to the cluster to spread out job execution and data handling loads to increase performance. With VA-based backup and restore, the vRanger server acts like the conductor of an orchestra: Once the administrator defines the plan – what to backup and when, and where to stream the backups – the VAs do all the work. The vRanger server also centrally monitors, alerts, and reports on all the action across all deployed VAs.
The data path with VA-based backup and restore is similar to when the vRanger server is installed in a VM – it’s very fast since it uses the internal host data path. The VA will default to use the SCSI HotAdd advanced transport method first, and failover to the Network transport mode if necessary. Data handling and data path options are user-configurable, so data handling can be performed by the vRanger server instead of the VA if desired, and specific data paths can be enforced by disabling automated transport failover.
The steps in VA-based backups involving the SCSI HotAdd advanced transport method can be broken down as follows:
The vRanger VA’s can process a large amount of data simultaneously, including backups, replications, and restores. This means if you need to restore a file or VM, you don’t have to interrupt an existing job (or wait until it finishes) to perform the restore. The maximum number of concurrent tasks running from each VA is eight, provided the VA is configured with enough CPU and memory. As a guideline, allow for two concurrent jobs per vCPU. Note that all VA vCPUs will be used for job-related activities. When the vRanger server is installed in a VM, one of the vCPUs will be used for the vRanger server and all other vCPUs will be used for job-related activities.
We also recommend adding a second SCSI controller to the vRanger VA so that it can mount more protected disks at the same time during the HotAdd operation. The vRanger VA will mount as many disks as permitted by its resource configuration and that vSphere allows.
The vRanger VA-based backup and restore architecture represents a very simple and low cost yet highly scalable, and high performing, architecture option for your virtual environment. This represents another flexible option that is complementary to vRanger-server based data handling, to help you achieve the most efficient data protection architecture to meet your objectives.