How Agentless VMware Backup Works

Agentless VMware Backup reads virtual machine data through the VMware vStorage APIs for Data Protection (VADP) and writes it to the backup destination. The read is performed by a virtualization proxy server, so neither the protected VMs nor the ESXi servers run any Catalogic DPX software.

Agentless VMware Backup Node Types

In addition to the Catalogic DPX Master Server, the following types of nodes are involved in Agentless VMware Backup.

• Proxy servers
• VMware vCenters
• VMware ESXi servers managed by VMware vCenter
• VMs
• Catalogic vStor or the NetApp storage systems (STORAGE_CTL nodes)

Agentless VMware Backup does not require the Catalogic DPX installed on vCenter, hosts, and VMs. The following table summarizes requirements for installation or configuration on the different types of nodes:

Node	Requires Catalogic DPX Installation	Requires Adding to the Enterprise
Virtualization proxy servers	Yes, with Virtualization Proxy Server installation setup type selected.	Yes
VMware vCenter nodes	No	Yes
ESXi servers	No	No
VMs	No	No
Catalogic vStor Server or NetApp storage system nodes	No	Yes

Backup Architecture: SAN, LAN (NBD), and HotAdd

Backup over a SAN or VMware HotAdd is recommended for superior performance. SAN has significant advantages over LAN (NBD):

• Impact on ESXi server processing is reduced. On a LAN, the ESXi server needs to read VMDKs and transfer them to the proxy server. On a SAN, the proxy servers read directly from storage, bypassing ESXi servers.
• Network contention is reduced.
• Network transmission time is reduced.
• Fewer connections are required on the vSphere server.

For many small- to medium-sized companies, LANs are cost-effective. In such environments, Agentless VMware Backup can be easily deployed for LAN backup, though performance is constrained by the LAN.

For HotAdd transport mode, vStorage APIs use the SCSI hot-add functionality of the ESXi host to mount a virtual machine’s associated virtual disks to backup storage, such as the Catalogic vStor server, for backup jobs. Using HotAdd requires that SCSI disks be used.

DPX selects the transport mode per job, based on availability. If all the datastores of the VM are accessible to the proxy over the SAN, the SAN transport mode is used. If not, DPX checks for HotAdd and uses it if it is available. If neither SAN nor HotAdd is available, the LAN (NBD) transport mode is used.

Base and Incremental Backups

The first backup of a VM is a base backup that copies the data of its virtual disks. Subsequent runs are incremental and transfer only the blocks that VMware Change Block Tracking reports as changed since the previous run.

For base backups, in most cases only allocated blocks are transferred. In certain cases, however, VMware does not generate the list of allocated blocks in a VMDK, and the backup transfers every block in the VMDK, including unallocated blocks. This happens when:

• The VMDK is not in a VMFS datastore.
• The VMDK is a virtual compatibility RDM disk.
• The VMDK is created as a result of a VM clone operation.
• The VMDK is created using vmkfstools in the eagerzeroedthick format.
• The VMDK is formatted using Long Format settings.
• The VM has one or more snapshots taken before DPX enabled Change Block Tracking for the backup.

Incremental backups are not affected by these constraints. For incremental backups, only changed blocks are transferred in all cases.

See also.

For the VMware limitation behind this behavior, read the VMware knowledge base article Changed Block Tracking (CBT) on virtual machines.

Data Reduction: Deduplication and Compression

Agentless VMware and Microsoft Hyper-V backups combined with Catalogic vStor Server or the NetApp storage systems provide powerful data reduction capabilities.

Catalogic vStor server provides data reduction through deduplication and compression (target-side data reduction). The deduplication and data compression capabilities for the Catalogic vStor Server and NetApp storage systems can be applied to the stored backup data. For more information on the effect of data compression, see Best Practices for Agentless VMware Backup.

Backup Considerations

• Change Block Tracking must be enabled in the VM. If change tracking is not enabled, DPX enables it before proceeding with the backup. If change tracking is disabled in the VM between backups, DPX enables it again for the next backup.
• A snapshot is not the same as a native VMware snapshot. A snapshot appears as a backup instance in the management interfaces and is the source for a restore. A native VMware snapshot is temporary and does not appear in the management interfaces.
• If you change a VM name after a backup is defined and run, the effect depends on the level of the VMware object selected for backup. If the VM is explicitly selected in the backup job, the backup job fails. If a higher-level VMware object, such as a datacenter, is selected for backup, the backup succeeds and stays incremental despite the changed VM name.
• Agentless VMware Backup skips VMs that have the Catalogic DPX Client installed only if VMware Tools was installed on the node before Catalogic DPX, and the corresponding job option is set. Otherwise, Agentless VMware Backup backs the node up. See Job Source Options for Agentless VMware Backup.

Attention!

If you perform storage migration of a VM in power-off mode, you must power on the VM at least once before a backup. Otherwise, the backup fails.

VM Exclusions from Backup

VMs are automatically excluded from backup with Agentless VMware Backup in the following cases:

• VMDKs shared across multiple VMs are not supported for protection due to VMware snapshot limitations.
• VMs are running a virtual hardware release earlier than 7. These early versions do not support Change Block Tracking.
• VMs with all VMDKs in an unsupported format such as physical compatibility VMDKs and independent disks. If the VM also has supported disks, they are backed up, and the unsupported disks are excluded. The VM is completely excluded from backup only if all the disks are unsupported.

Additionally, if you install DPX in a VM for application-aware protection, or for some other reason, you can set the Agentless VMware Backup to skip that VM.

Excluded VMs can be backed up with Block backup if DPX is installed on the VM.

When VMs and VMDKs are excluded from the backup, the reported job status depends on the Report Job Failed if VM/VMDK is Skipped job option. See Job Source Options for Agentless VMware Backup.

See also.

For a complete list of VMware snapshot limitations, see the vSphere Virtual Machine Administration guide.