Does anyone know why, when backing up a VMWare VM with Changed Block Tracking enabled, if that VM is powered off then MSP360 Backup reads the entire disk and ignores the fact that no bytes have changed according to CBT. I have several VMs that I just use intermittently. If they are powered on then backups are usually fairly quick but when they get switched off they can take many hours or even days.
Spending eight hours to read a 100GB disk on a remote system over the internet just to decide that nothing has changed since last time seems very inefficient when it could, if CBT is showing no changes, just look at the last update time of the disk and realise that it's not been used at all since the last backup.
To confirm this I have run the following tests. Note that the VM has CBT disabled and re-enabled before the tests to ensure it's clean and not corrupted.
1) VM powered off - Very slow backup. 0 CBT bytes, entire disk read over network, 0 bytes to backup.
2) VM powered on - Faster backup, CBT changes found, chunk of disk read over network, smaller chunk backed up.
3) VM powered on - Fast backup, fewer CBT changes found, smaller chunk of disk read and backed up.
4) VM powered off - Fast backup, few CBT changes found, small chunk read and backed up
5) VM powered off - Very slow backup. 0 CBT bytes, entire disk read over network, 0 bytes to backup.
So a second or subsequent backup of a powered off VM results in horrendous performance. There must be a way to optimise this to backup powered off VM s much faster. One of the reasons I switch them off is to reduce resources and consumption on the hosting server but MSP360 then goes and pointlessly thrashes the disk for hours or days on end for no good reason.
Any suggestions how to work around this until it gets a proper fix in the application?
In backup & recovery there’s something known as the 3–2–1 Rule. This refers to having 3 copies of your data, on 2 types of media, with (minimum) 1 copy offsite.
If your company is saving one copy in the same physical room, this isn’t an entirely abnormal practice. The better question is, where is/are the other copy/copies?
Thanks but I'm not sure if that's relevant. I have several backup plans going to different onsite and offsite locations but used a single one as the example to simplify it. The main point is that backing up a powered off VM reads the whole disk when the operation could be massively shortcut by just checking the last modification times.