Just curious: Processed/Read/Transfered question

[KOPFteam]

I don't understand why 2 simple Replication jobs behave differently:

We have 2 ESX and Veeam physically installed on a 3rd server.

• Job 1 is replicating one VM from ESX1 to ESX2
• Job 2 is replicating one VM from ESX2 to ESX1

Result:

• Job1: Processed 420GB (100%); Read 410GB; Transfered 670MB; Duration 1:14; backup proxy on target ESX (hotadd); target proxy the VM itself on the source ESX (hotadd)
• Job2: Processed 848GB (100%); Read 656MB; Transfered 387MB; Duration 0:05; backup proxy on target ESX (hotadd); target proxy on target ESX (hotadd)

The amount of data that was transfered is OK, but why does Job1 read the whole disk for more than a hour and Job2 just reads 600MB from a much bigger disk in a few minutes? Changed block tracking is enabled on both jobs. Proxy settings are set to "automatic selection" on both jobs. Both jobs successfully ran a few hours before (with the same behaviour).

Actually, this is not a real problem but I want to understand what's going on.

PS: This question is not related to my urgent, still unsolved "Failback failed" post I posted last friday.

Florian Pürner
KOPFteam GmbH
Munich, Germany

[Vitaliy S.]

Florian,

What are the bottlenecks reported for each job? Do they run at the same time?

Thanks!

[KOPFteam]

Hello Vitaliy,

thanks for your answer. No, the jobs are running sequentially. The bottlenecks are:

Job1: Source 98% > Proxy 26% > Network 0% > Target 3%; Primary bottleneck: Source
Job2: Source 85% > Proxy 25% > Network 19% > Target 70%; Primary bottleneck: Source

Florian

[Vitaliy S.]

Hmm... what are the processing modes effectively utilized by each job? NBD or hotadd?

[KOPFteam]

Both jobs use hotadd to add the disks to the proxies.

Both jobs report "Changed block tracking is enabled" but
The first job (the one that reads the whole disk) reports
Festplatte 1 (420.0 GB) 410.0 GB read at 101 MB/s
The second job (the one that only reads a few MB) reports
Festplatte 2 (768.0 GB) 141.0 MB read at 38 MB/s [CBT]

So the first job doesn't use CBT for this (the only) disk. Any way to find out why?
The virtual hardware version of this VM is 7 and there are no snapshots. CBT is enabled in the advanced job settings.
The job is using the VM itself as target proxy. Maybe this is the reason?

Thanks,
Florian

[foggy]

Right, if the proxy is used to hotadd disks of other VMs, CBT cannot be used during backup of this proxy.

[jcurtiss7492]

I have (1) stand-alone physical server for Veeam doing all roles. Additionally, I have installed a backup proxy role on one of the VM's.

I'm getting a line in my backup logs for one of my VMs in a backup job that states, "Changed block tracking has been disabled for backup proxy VM". However, there are all green check marks, no red. Does this indicate a problem? Things I have checked:

1. the setting is checked to "Use changed block tracking data" and "Enable CBT for all protected VMs automatically".
2. the hardware version of the VM is ESXi 6.0 and later (VM version 11)
3. the other 4 VMs in this backup job are using changed block tracking ok and don't have this line in their log
4. I do have a backup proxy VM but it is a different VM and not the server with the issue in the log

Is this a cause for concern? Thanks in advance!

[Vitaliy S.]

Hi Jeff,

Job shows all green, because CBT cannot be used for hotadd backup proxy servers. This is an expected behavior, please check out the post from Foggy for more details.

Thanks!