Building a VMware vSphere proxy is a straight forward process that does not require any particular special care: it is a matter of few clicks in the Veeam Backup & Replication UI to add it.
However, in certain circumstances there might be some additional tweaks one may take into account to optimize the infrastructure.
Physical proxy can be used with two different transport modes: DirectSAN and Backup from Storage Snapshots (BfSS). From an infrastructure point of view, the difference is mainly related to the way the proxy interacts with the source storage array.
In the Direct SAN method, the proxy must be zoned with the source storage and the LUNs containing the virtual machines to be protected have to be mapped, whereas with the BfSS only the zoning is required.
With that being said, it is easy to understand that there might be some risks related to the Direct SAN: since the LUNs containing the source VM are visible from the disk manager of the proxy and, in some cases, the proxy is also used as repository server, managing both local and SAN volume may be prone to human mistakes. Eliminating the risk of creating a new file system on the wrong volume (meaning the vSphere datastore) becomes key.
When deploying the datamover package, Veeam Backup & Replication disables both the Windows automount feature and remove the letter assigned to previously mounted drives (automount scrub) automatically. The volume is still mounted with read-write access.
From a backup process, point of view of the read-write access is not required and
DISKPART utility can be used to force a different access method and set the LUNs where the source VMs are running to read-only:
DISKPART> list disk DISKPART> select disk x DISKPART> attributes disk set readonly DISKPART> attributes disk
The process has to be repeated for all the LUNs where a VMware vSphere datastore is created.
On the other hand, BfSS does not require this process since only the storage array’s snapshots are presented to the proxy. During the backup operations, a new volume will be visible from the Windows Disk Manager but it will never be the source LUN.
Data restore through the SAN is possible only for thick VMDKs and usage of this mode forces Veeam Backup & Replication to restore in thick format.
In order to make the restore work properly, the target volume must be set back to read-write. Use DISKPART as follow to change the policy:
DISKPART> list disk DISKPART> select disk x DISKPART> attributes disk clear readonly
Remember that several factors can negatively affect backup resource consumption and speed:
Compression level - It is not recommended to set it to “High” (as it needs 2 CPU Cores per proxy task) or to Extreme (which needs a lot of CPU power but provides only 2-10% additional space saving). However, if you have a lot of free CPU resources during the backup time window, you can consider to use “High” compression mode.
Block size - The smaller the block size, the more RAM is needed for deduplication. For example, you will see a increase in RAM consumption when using LAN mode compared to Local Target, and even higher RAM load (2-4 times) when using WAN. Best practice for most environments is to use default job settings (Local for backup jobs and LAN for replication jobs) when no other is mentioned in the documentation or this guide for specific cases.
Antivirus - see the corresponding KB1999 for the complete list of paths that need to be excluded from antivirus scanning
Third party applications – it is not recommended to use an application server as a backup proxy.