Consolidated Backup w/ BExec 11d planning

[Aszurom]

I'm currently using Backup Exec 11d with agents deployed into my VMs to do my nightly backups. I'd much rather back the whole .vmdk to make restoring a disaster-fied machine that much easier.

I want to talk to some people who are already doing this and get an idea of what gotcha's lie ahead, and what the current method of doing this is. I'm low-to-middling experience with VMware ESX, but zero experience in consolidated backup operations.

Here's my current layout:

2 ESX 3.0.1 servers, an EMC SAN, Virtual Center hosted on a physical 2003 server, and another 2003 server running backup exec that happens to have a scsi direct-attached 300 gig drive shelf available on it.

I want to get VCB working before I push the servers up to 3.0.2 or 3.5.

VirtualInfrastructure is already 2.5.

?) Where on vmware's website do they hide the Backup Exec 11d scripts for VCB intergration? I found a post from sept 2007 on the forum that linked to a set of scripts, but I'm dismayed I couldn't find it on the site - and want to make sure I've got the latest version of it. Please don't link direct to the file, but clue me on where they keep it because it's either hard to find or I'm blind.

?) The documentation I'm reading says that you put "<vcb-path>\backupexec\pre-command.bat <jobname> <VMs>" for the pre-script. I'm not seeing where you differentiate if you want to copy the .vmdk over to the VCB server and then have backup exec back that up from there, or mount the vmdk as a folder in the filesystem and do a file-level backup. This seems too simple to be all there is to it. What am I missing here?

?) Since I have 300 gig of space to back vmdk files up in on VCB machine but all my VMs won't fit in there at once, I'd have to do a few VMs at a time, and then flush those out and start a new job on another group? Would I have multiple backup exec jobs that - for example - back up VM1, 2, 3 and then finish, start another backup exec job that appends to the media from the first job, backs up VM4, 5, 6, and so on?

?) Is VCB and Backup Exec running on the same physical machine the preferred arrangement?

[jfields]

Aszurom,

I can answer a couple of these questions for you:

Where on vmware's website do they hide the Backup Exec 11d scripts for VCB intergration?

They are actually in the Drivers and Tools section of the VI downloads http://www.vmware.com/download/vi/drivers_tools.html.

Is VCB and Backup Exec running on the same physical machine the preferred arrangement?

That makes the most sense. Check out this doc: http://entsearch.symantec.com/search?p=R&srid=S10%2d3&lbc=symantec&w=vmware%20faq&url=http%3a%2f%2ff...

Also this: http://seer.support.veritas.com/docs/288196.htm

[dmorgan]

I want to talk to some people who are already doing this and get an idea of what gotcha's lie ahead, and what the current method of doing this is. I'm low-to-middling experience with VMware ESX, but zero experience in consolidated backup operations.

Gotcha's are many, but once it works it is fairly simple. First and foremost, don't just install VCB and then go right to pre- and post- scripts in BackupExec and wonder why nothing works. Do it step by step. First, get VCB installed on your proxy server. You didn't mention if your backup server had SAN access or not. If it has Fibre-HBA or iSCSI access to the SAN, then great, make that your proxy server, as the 300GB local SCSI storage should be adequate for fullvm backups. Otherwise, you will need to build a proxy server (physical or VM) that has access to the datastores that you need to back up. Next, learn the VCB commands from the command line. Start with VCBName to see if you have set everything up properly, and can see the VMFS partition(s) and the VM's that reside there. Once this works, move on to vcbmounter.exe and try some file-level mounts of a windows machine. This is faster than a fullvm mount, as no data has to be copied to the proxy. If you can do this, then try a vcbmounter.exe command to do a fullvm mount of a small VM. Small, because it has to copy all of the associated files from the VMFS partition to the proxy, and if you do a test on a VM with some monster .vmdk's you will have to wait for quite some time potentially. If you get to this point, and his all works, move on to the pre and post commands, but DO NOT run them from within backup exec. Instead, run them manually, and tweak them until they do what they are supposed to do. Once these work this way, then work on using backup exec to run the pre and post commands. From experience, everyone installs everything and then skips right to their tape backup software and scripts, and runs them and nothing happens. Then comes to post, why isn't it working?!? Take it step by step, and you will not only have an easier time, you will learn what you need to know about VCB along the way.

?) Where on vmware's website do they hide the Backup Exec 11d scripts for VCB intergration? I found a post from sept 2007 on the forum that linked to a set of scripts, but I'm dismayed I couldn't find it on the site - and want to make sure I've got the latest version of it. Please don't link direct to the file, but clue me on where they keep it because it's either hard to find or I'm blind.

Not really sure... I believe that another poster pointed towards this location. We settled on Arcserve, mainly because their latest version fully integrates with VCB, no need for pre and post scripts anymore, however I have tested Backup Exec, Arcserve, and several other backup platforms with VCB and have made them all work.

?) The documentation I'm reading says that you put "<vcb-path>\backupexec\pre-command.bat <jobname> <VMs>" for the pre-script. I'm not seeing where you differentiate if you want to copy the .vmdk over to the VCB server and then have backup exec back that up from there, or mount the vmdk as a folder in the filesystem and do a file-level backup. This seems too simple to be all there is to it. What am I missing here?

This is all determined by the vcbmounter.exe command. While when you use the pre and post scripts you don't directly have to use these, the scripts are calling this executable and plugging in the variables. This is why I recommended you start from scratch and learn these commands first, they aren't that complicated. If you get stuck at any point, simply post to this group and we can point you in the right direction.

Examples:

vcbvmname -h pcgvm1.parkcitygroup.local -u root -p ******* -s any vcbvmname -h pcgvc.parkcitygroup.local -u administrator -p ******* -s any

Two commands above, both do the same thing. One goes directly to a single ESX server, the other to the VC server. Fairly straight forward.

vcbmounter -h pcgvc.parkcitygroup.local -u administrator -p ******* -a name:"ProphetServer" -r d:\mnt1\ProphetServer-fullVM -t fullvm vcbmounter -h pcgvc.parkcitygroup.local -u administrator -p ******* -U d:\mnt1\ProphetServer-fullVM

The two commands above mount, and unmount, a fullvm. Notice the -t fullvm switch. This is what tells VCB to do a fullvm backup and not a file level backup. The rest are fairly self explainitory.

vcbmounter -h pcgvc.parkcitygroup.local -u administrator -p ******* -a name:"ProphetServer" -r d:\mnt1\ProphetServer-file -t file vcbmounter -h pcgvc.parkcitygroup.local -u administrator -p ******* -U d:\mnt1\ProphetServer-file

Same as the above commands for a fullvm backup, but using -t file, so it does a file level backup of a Windows VM.

?) Since I have 300 gig of space to back vmdk files up in on VCB machine but all my VMs won't fit in there at once, I'd have to do a few VMs at a time, and then flush those out and start a new job on another group? Would I have multiple backup exec jobs that - for example - back up VM1, 2, 3 and then finish, start another backup exec job that appends to the media from the first job, backs up VM4, 5, 6, and so on?

We have a similar situation. I have about 2TB of VM's to back up, so when we do fullvm backups I don't have 2TB of storage on our proxy/backup server. Simple solution, is to group your VM's into smaller groups, or run individual jobs for each VM you back up. All you REALLY need is enough local storage to fit your single largest VM. So long as you have that, you can use VCB to back up everything, individually or in groups.

?) Is VCB and Backup Exec running on the same physical machine the preferred arrangement?

See above. We do ours that way, Arcserve and VCB on the same physical machine. This is the easiest method, although by no means the only way to do it. Often times your proxy must exist on a separate machine, because your tape backup server doesnt have fiber connectivity or whatever to the SAN. Your proxy MUST see the vmfs LUN(s) on the SAN by the same name/path as your ESX servers do. Thus, it is your choice, however it is often simpler if they can be on the same machine. Otherwise, you just make sure the proxy can connect via VCB as described above, and then back up the proxy with backup exec.

If you found this or any other post helpful please consider the use of the Helpfull/Correct buttons to award points