Type of backup job for quick recovery

[eversys]

We have a VMware ESXi 5.5 host. We would like to backup all VMs to a local storage server. In case of disaster - i.e. the VMware host is unavailable - then we would like to be able to either mount the backup files or import them into a newly installed VMware Host in order to get the infrastructure up-and-running as quickly as possible.

Which type of Veeam backup job would best suit our needs: Backup, Backup Copy or VM Copy?

Thanks for your help.

[Shestakov]

Hello,

Backup jobs are the best to fulfill your requirements.
Backup copy job is made to achieve additional reliability by keeping another copy of VMs` backups offsite.
VM copying can be helpful if you want to move your datacenter, create a test lab and so on.
However for the fastest recovery of the most critical applications it`s recommended to make replication jobs of those VMs.

If you told me about your backup policy and infrastructure I would be able to give your more advice. Thanks.

[eversys]

Thanks for your advise.

Up until now we used Symantec System Recovery to create image backups of the native OS from within each VM. Due to time-consuming restore of a complete system back to VMware (the native Symantec backup file needs to be converted and then restored), we decided to move to Veeam!

We have 1 x VMware host (with approx 6 VMs - mostly Windows Servers including SBS 2011 & SQL) and 1 x Windows Server 2012 R2 as a backup file store, attached to a LTO6 autoloader. Basically, we need a disk to disk to tape strategie, which also enables quick VM recovery (in a real disaster - VMware host dead), along with file (and mail) restore (the normal request from end users). The replication job you mention, requires a second VMware host/datastore - right?

Thanks for your help here.

[Shestakov]

You are welcome. Glad you are with us now!

It`s great to have both disk and tape repositories for backup storage. So, you can make backup jobs to store backup files on the disk repository + backup copy jobs to copy them to the tape repository.
Is tape repository also placed in the different site?

Veeam provides a number of data recovery tools including those you`ve described. Also there are application-specific Veeam Explorers for item-level recovery.

The replication job you mention, requires a second VMware host/datastore - right?

Correct.

[eversys]

We have setup a disk repository - however, for the tape backup we did not create a tape repository but a "Backup t Tape" job for the disk repository and a "File to Tape" job for local files on our storage server. Is that OK? We only have 1 site.

[Shestakov]

Is it ok or not depends on your resources and needs.
The best practice is to have three copies of your: production data, backup and its copy on two types of media(e.g. disk and tape) and to have one of the copies offsite(another site or cloud). It`s called 3-2-1 backup rule.
Your case on the whole looks good. To increase reliability I would suggest to move tape storage offsite or to add cloud storage. It will protect your data from the case of disaster(fire, flood, etc.)

[eversys]

We do have offsite backups... although the tape device is onsite, the tapes are kept offsite.

With our old backup strategie (SSR), we ran full system backups (images) everyday to disk. The images where transferred to tape every weekend. Data was written to tape everyday. In Veeam, I am a little confused as to the backup mode I should be performing:

1. For my disk-based backup, I thought about setting up a "Backup job" of all of my VMs, with 6 restore points (Rentention policy) using a "Reversed incremental" Backup mode and an "Active full backup" each Sunday.
2. For my tape backup, I thought about setting up a "Backup to Tape" job and selecting my disk "Backup Repository" which I created in step 1. This would be run everyday.

What are your thoughts here?

Another question... is it best practise to have separate Backup jobs for each VM and have them run after another or is it perfectly OK to have all VMs backed up in 1 job?

[Shestakov]

For my disk-based backup, I thought about setting up a "Backup job" of all of my VMs, with 6 restore points (Rentention policy) using a "Reversed incremental" Backup mode and an "Active full backup" each Sunday.

It should work great.
You can also consider forever forward incremental backup. Here is a forum discussion describing advantages of the method.

For my tape backup, I thought about setting up a "Backup to Tape" job and selecting my disk "Backup Repository" which I created in step 1. This would be run everyday.

You can backup not the disk, but restore points of the original backups. For that purpose backup copy job is made. Please get familiar.

Another question... is it best practise to have separate Backup jobs for each VM and have them run after another or is it perfectly OK to have all VMs backed up in 1 job?

It`s ok to have several VMs in one job.

[foggy]

1. For my disk-based backup, I thought about setting up a "Backup job" of all of my VMs, with 6 restore points (Rentention policy) using a "Reversed incremental" Backup mode and an "Active full backup" each Sunday.

Weekly active fulls for reversed incremental is a bit overkill, please review this thread for recommendations on full backup frequency: How often do I need a full backup?

is it best practise to have separate Backup jobs for each VM and have them run after another or is it perfectly OK to have all VMs backed up in 1 job?

Actually, it is recommended to combine VMs created from the same template (running the same OS, the same apps) into one job to get better dedupe rate between them.

[eversys]

@Shestakov

You can backup not the disk, but restore points of the original backups. For that purpose backup copy job is made.

I would like to backup to tape. I was unable to create a tape repository, hence, I have no "Backup repository" for my "Backup copy job". That's why I decided to use a "Backup to Tape" job under "Tape Infrastructure". Here it allows me to either select a "Backup job" or "Backup repository" - hence, I thought that my local backup repository was the correct choice. Please correct me if I am wrong.

@foggy

Weekly active fulls for reversed incremental is a bit overkill.

I am still a little unclear as to which backup mode I need to select in order to have a weekly full (on Sunday) and daily incremential (Monday thru Saturday) as described in the article "Active Full Backup": http://helpcenter.veeam.com/backup/70/v ... ackup.html. In order to achieve this I thought I needed to set an "Active Full". What I don't understand, if I just select "Reversed Incremental" then when do I get my full backup (only the first time I run the job?)???

combine VMs created from the same template (running the same OS, the same apps) into one job

I have a combination of Windows Server 2003, 2008 & 2012 to backup, all have different roles and programs. Should I still combine them into 1 job?

[veremin]

What I don't understand, if I just select "Reversed Incremental" then when do I get my full backup (only the first time I run the job?)???

With the reversed incremental mode, the latest restore point is always a full backup and the changes occurred are inserted inside this .vbk file. With reversed incremental mode in place, there is no need to run active full backup so regularly - once a month or so would be enough. Check our user guide for more information regarding backup modes.

Thanks.

[Shestakov]

I see your point. You are correct, you can choose both backup jobs and backup repositories as a source of backup to tape job. That should work fine.
Why do you need to do active full weekly?
I would recommend you once again to consider Forever Forward Incremental backup method rather than reversed one, since it`s more fitted to tape infrastructure.
You are correct, with reversed incremental active full is made only with the first run.

I have a combination of Windows Server 2003, 2008 & 2012 to backup, all have different roles and programs. Should I still combine them into 1 job?

When you combine several VMs in 1 job, similar data blocks can be deduplicated, so it can decrease ultimate size of the backup file. So it makes sense to combine them.
Thanks.

[eversys]

Thanks for all your feedback.

Last night my first backup to disk of my VMs ran. Here the result:

Looks good and I am very surprised with the backup size! Just for clarification, could I 'mount' any of these VMs from disk (using Instant VM recovery?) and have them run in a disaster scenario?

What I now want to do is get these VMs to tape. I have setup a "Backup to Tape" job and selected the above VMWare backup job, created a media pools of tapes and have it run automatically after the above job finishes. Is this the correct way?

What I also want to know is what I would do in a real disaster scenario when no servers exist and I only have a tape backup of the VMs?

Finally, what we also have are file shares on our "backup server" containing the LTO6 autoloader. We have created a separate "File to Tape" backup job to have our data backed up to tape. I have this job run at the same time as my first backup job of the VMs to local disk. Is this OK. Unfortunately, I couldn't specify that the job runs after my other jobs.

Thanks for your continued help.

[EDIT]
Not sure why my backup screenshot doen't show in my previous post. Here are the results again:

[Shestakov]

VBR uses the Veeam vPower technology to mount a VM image to an ESX(i) host directly from a compressed and deduplicated backup file. Since there is no need to extract the VM from the backup file and copy it to production storage, you can restart a VM from any restore point (incremental or full) in a matter of minutes.
Please check prerequisites for Instant VM Recovery.

What I now want to do is get these VMs to tape. I have setup a "Backup to Tape" job and selected the above VMWare backup job, created a media pools of tapes and have it run automatically after the above job finishes. Is this the correct way?

Yes, it`s.

What I also want to know is what I would do in a real disaster scenario when no servers exist and I only have a tape backup of the VMs?

You would need to add an ESX(i) host and tapes to VBR and make the recovery.
Thanks.

P.S. Your screenshot is posted fine.