Details on how VM backups work

[Thernlund]

I've been looking through technical document for a couple days and I'm either not seeing or not understanding some technical details concerning VMWare backups.  Sme of the document are long winded and not particularly direct in some details, so it's certainly possible I read it and missed it.  

First my setup...

• Backup Exec 2014
• Windows 2008 R2
• ESX 5.5
• EMC SAN stores the VMs
• About 10 VMs are being backed up, the largest of which is 1.5TB.
• I backup through the vCenter server.
• I am NOT using GRT on any of them as I do not have the disk space for staging.
• I backup directly to LTO tape (see previous bullet re: disk space).

What I'd like to know is the following...

1. How are VMs backed up exactly?  Snapshots?  Or direct block reads?
2. Same question as #1, full vs. incremental?

I'm concerned because the SAN is nearly full.  There's some wiggle room, but nowhere near enough to create a snapshot of several of the larger VMs.  A full snapshot of a 1.5TB VM would fail.  I have maybe 300GB of empty space on a 4TB SAN.

I do have a basic understanding of how VMWare snapshots work, but certainly I'm not all-knowing.  I could certainly be missing on the front.

Any knowledge on these topic where backups is concerned is appreciated.

[pkh · Accepted solution]

A snapshot is taken, then blocks are read. This applies to both full and incremental backups

[Thernlund]

Since I wrote this I found some other 3rd-party documentation that describes the process a bit more succintly.  The gist is that backups don't really take long enough to allow the snapshot to grow very much, especially when done during non-business hours.

The kink in my understanding was on the VMWare side concerning how snapshots work, not the BE side.

Thanks for the reply.

[Andrew_Madsen · Accepted solution]

Look at these two articles:

NetBackup 7.6 Blueprints - VMware

NetBackup 7.6 Blueprints - Accelerator for VMware