After splitting one job into two, overall backup is larger

[Tijz]

Hi,

I had one backup job for all my vSphere virtual machines (41).
I've set compression to "none" and optimize for to "local target". (I'm using a deduplication appliance)

A full backup is about 4,5TB
For months daily incrementals were about 240GB.

Now I split the job into two jobs. One job backs up 19 and the other 22 VM's.
A full backup of each job is about 2.25TB, so 4.5TB together

But the daily incremental backups amount to about 400GB (200GB each) !! Almost twice as large as when I was using just one job.
How is this possible?

I'm using Veeam B&R 6.1.0.181 with vSphere 5.0.0
I'm backing up directly from an iSCSI SAN.

Thanks.

[foggy]

Mattijs, Veeam B&R performs inline data deduplication on the job level (not between different jobs). Since you've split your job into several ones, data is not deduped to the same level anymore.

[Tijz]

Oh yes you might be right. Allthough I disbabled compression and optimized for "local target" I still checked "inline deduplication".

but.... How do I determine the amount of deduped data then? I assume it's the difference between the amount of data "read" and "transferred" (when you look in the job statistics).
For both backup jobs the data "read" and "transferred" are about the same, so, as you said, no dedupe (anymore?).
But the amount "read" of both jobs together is way more then when I had just one job.

So now:
Job 1 "read" is 208GB
Job 1 "Transferred" is 206GB

Job 2 "read" is 185GB
Job 2 "transferred" is 166GB

Together about 394GB is "read".

When I had just one job it was like this:
Job "read" was 240GB
Job "transferred" was 219GB.

So still a lot less than it is now. I don't think this is due to deduplication, do you?
Or is deduplication being performed earlier?

[Gostev]

Yes, deduplication is performed both at the source and at the target - just on the different scopes of data.