Enhancement request

[Spex]

At the moment I try to find the best way to set up scheduling for about a virtual center with 1000 vm's.
Storage target (LAN) is a dedupe appliance. ESX storage over Fibrechannel. 700 of 1000 vms are should be backuped.
One of my goals is to add new vm's (semi-) automatic to existing backup jobs.

Setting up scheduling per vm is flexible but very, very timeconsuming. Sure you can use powershell but coordinating starttimes for 700 job's isn't funny. Also no automatic.

Jobs per datastore have 2 problems: we use storagemigration and some of the vm's that need not to be backuped reside on the same datastores as the others. Datastores with 2 TByte are a bit large for one job (synthetic backup times....)

We do no want to restructure the virtual center for better job granuarity. The current structure has 2 purpose's: resourcepools and security (organsiational structure)

Per esx server similar problems as datstores.

Since Veeam handels already customized attributes in virtual center why not use them?
The idea is: each vm has an attribute veeamjob with following content: job1|job2|job3|...|none
Within Veeam you define jobs job1,job2,job3... and let them backup the according vm's.

Wouldn't that be nice?

[foggy]

Stefan, actually you could use any containers (such as folders or resource pools that are not changed during vMotion) to select VMs for backup in your job. Wouldn't that resolve your issues?

Also I would note that you should not coordinate the jobs start times in Veeam B&R v6 as it handles this automatically allowing you to limit the number of concurrently running jobs per proxy/repository.

[tsightler]

I have a Powershell script that effectively does this today but agree that this would be nice functionality. In the interim I have attached my script to this post. It would need to be modified for your use case, but should be pretty simple. The script basically reads a customer attribute "VeeamJob" from all VMs in vCenter and assigns the VMs to jobs with the same name in Veeam. If the attribute is changed/removed the VM will be removed from the job. If the job does not exist it will be created with some default parameters.

[Spex]

Stefan, actually you could use any containers (such as folders or resource pools that are not changed during vMotion) to select VMs for backup in your job. Wouldn't that resolve your issues?

Also I would note that you should not coordinate the jobs start times in Veeam B&R v6 as it handles this automatically allowing you to limit the number of concurrently running jobs per proxy/repository.

Our actual folder/resourcepool design isn't suited for job scheduling: e.g. one resourcepool "Production" with 300 vm's or regarding our folder design: 50 ou's/folders - one of them with 1 vm the other with 120 vm's...

[Spex]

I have a Powershell script that effectively does this today but agree that this would be nice functionality. In the interim I have attached my script to this post. It would need to be modified for your use case, but should be pretty simple. The script basically reads a customer attribute "VeeamJob" from all VMs in vCenter and assigns the VMs to jobs with the same name in Veeam. If the attribute is changed/removed the VM will be removed from the job. If the job does not exist it will be created with some default parameters.

very nice, I will have a very closer look at your script when it comes to final design. Thanks a lot for sharing.

[dellock6]

If it's about performances, in each job VM are processed sequentially, so there is no difference in having 10 or 1000 VMs in a job. Cuncurrency is managed with several jobs, not inside the same job.

[domenicalvaro]

Our actual folder/resourcepool design isn't suited for job scheduling: e.g. one resourcepool "Production" with 300 vm's or regarding our folder design: 50 ou's/folders - one of them with 1 vm the other with 120 vm's...

I understand why Resource Pool may not be suitable for backups, but surely the VM folder structure provides the best way to arrange VMs so that you can easily back them up in a logical order. I realise that I don't know your specific setup, but to me it would make sense to re-organise the folders to acheive your goal.

[tsightler]

I understand why Resource Pool may not be suitable for backups, but surely the VM folder structure provides the best way to arrange VMs so that you can easily back them up in a logical order. I realise that I don't know your specific setup, but to me it would make sense to re-organise the folders to acheive your goal.

In many organizations the folder structure is already used for security related access to VMs (I believe that the OP may have stated that his organization uses them in this way). In many cases this is "non-ideal" for backup and cannot easily be changed so your left with a non-optimal setup. Using attributes provides maximum flexibility in that they are not dependent on anything else in the infrastructure and can easily be automated with additional scripting.

[Spex]

In many organizations the folder structure is already used for security related access to VMs (I believe that the OP may have stated that his organization uses them in this way). In many cases this is "non-ideal" for backup and cannot easily be changed so your left with a non-optimal setup. Using attributes provides maximum flexibility in that they are not dependent on anything else in the infrastructure and can easily be automated with additional scripting.

I can only agree with tsighter. The vm folder structure's primary purpose is to organize the access to the vm's. If the structure also fits for backup scheduling you are lucky. What we need is an independent mechanism.