Job creation planning

[BuffaloBucks]

There are 14 LUNs attached to our 4 node 2016 Hyper-V Cluster and I am planning to use storage snapshots to create backups for SQL Servers, Large File Servers and other Misc application servers. I'm not sure exactly sure of the best number of jobs to have as Veeam is much different than DPM that we have been using but I was thinking of having a job for SQL Servers, another job for the large file servers and another job for the Misc application servers.

Example: I have 20 SQL Server VMs in the "SQL_Job" being backed up using storage snapshots. I want to create a recovery point for one DB on one of the VMs (this VM DB server holds many DBs for many different application servers) before upgrading the application server.
1.) Can I go to that DB in the Veeam job and just create a recovery point for that one DB?
2.) Does it have to backup all servers in that job?
3.) Since I am using storage snapshots and this VM is on a different LUN than other VMs in this job, will it create a storage snapshot for each LUN in this job or just the LUN with this VM?

Regarding storage snapshots that are created by Veeam for backups, do these storage snapshots stay on the array until the recovery point expires or do they get deleted once the Veeam job completes and the data is transferred to the Veeam backup repository?

[Egor Yakovlev]

Hi Al,
- as mentioned in a parallel thread of yours, Storage snapshots are relevant for VMware Infrastructures only
- image level backup of a virtual machine server takes entire machine data by default. You can exclude specific VM disks from processing, though.
- alternatively, you can perform agent-based backup and backup particular guest OS volumes or files.
- in Hyper-V environment backup is done via VSS Snapshot, and it is done per-volume with said VM disks. Those snapshots are removed after backup.
Cheers!

[BuffaloBucks]

Yes, I should start using the correct terminology as I am understanding this better. I have a Hyper-V environment so it would be a Volume Shadowcopy Service snapshot. I haven't received confirmation yet, whether or not the hardware VSS provider from our storage vendor is supported with Veeam so I am still working on that. In the meantime, I have created a couple jobs and backed up a couple VMs successfully using the default Software VSS Provider. A couple things I have noticed and wondering about:

1.) I have an Off-Host backup proxy properly setup and zoned via FC to see all the Hyper-V LUNs following all the Veeam documentation, however when I run a Veeam job, the data is going over the 10Gb Ethernet instead of over Fibre Channel. I watched the NIC traffic stay sustained to about 500 Mbps during the backup. Why isn't it going over FC? Is it required to be using a Hardware VSS Provider to achieve LAN-free backups?

2.) While watching the VM in Hyper-V manager during the backup utilizing the Software VSS Provider, I see it create a snapshot/checkpoint of the specific VM and then delete it when the job finishes. Now comparing that with using a Hardware VSS Provider, does the Hardware VSS Provider take a snapshot of the entire CSV LUN, including VMs that are set to backup and VMs that are not going to be backed up? Does it keep this storage array snapshot on the array and visible only through the storage array user interface and copy just the files needed for the VM(s) that are set to be backed up or does it function differently?

Thanks.

[Egor Yakovlev]

Hi Al,
1.) without properly configured Off-Host proxy, Veeam will use default On-Host proxy instead, which sits directly on top of a Hyper-V host where VMs are being processed from. So data will flow SAN->Hyper-V host(Proxy)->Backup Repository.
2.) No matter of VSS Provider type(Hardware of Software), VSS snapshot takes entire volume snapshot, even for those VMs that are not included in Backup Job but are located on same volume. However, VM checkpoints are taken per-VM(according to application-consistency settings set under Job-Guest Processing tab) and only VMs that are set to be application consistent in backup job will have a checkpoint created. VMs that are not in the backup job are not touched, VM disk data is not processed, and are not affected in any way.
/Thanks!