Performace with Windows Server 2022 RCT vs. 2012 CBT

[check4backup]

Good evening,

I notice that the backup from a Hyper-V host with Server 2022 takes twice as long as from a Hyper-V host with Server 2012R2. However, the new Hyper-V host should be faster than the host with Windows Server 2012 R2.

As of Server 2016, the native RCT is used. Is it possible, or can someone confirm from experience, that backups with RCT take longer than with the Veeam CBT driver. Or do I need to look for another problem?

Strangely, only the forward incremental backups are slower. The full backups are not.

The full backups run at 500-700 MB/s, the forward incremental at 150-200 MB/s.

On the W2K12R2 host, all backups run at about 400-500 MB/s.

Is RCT slower?

The VM is a Domain-Controller (W2K22) with Application-Aware Processing (copy only).

Best regards

[HannesK]

Hello,

The VM is a Domain-Controller (W2K22) with Application-Aware Processing (copy only).

that sounds like, it's the same VM. How is the overall backup time and the change rate (data read) compared to earlier when running on Server 2012R2?

Best regards,
Hannes

[check4backup]

Hello,
sorry for the delay. No, they are diffrent VMs. It is quite possible that this is not a problem at all. Let's just share some information so it's clear how Veeam works and maybe this clears up.

Hyper-V Host W2K12 - VM (Domaincontroller) W2K12R2
Duration: 02:10 (incremental)
Processing rate: 491 MB/s
03:38:31 Creating snapshot Microsoft Software Shadow Copy provider 1.0 (mode: Veeam application-aware processing) 00:31
03:39:05 Hard disk 1 (70 GB) 17,9 GB read at 491 MB/s [CBT] 00:38

Hyper-V Host W2K22 - VM (Domaincontroller) W2K22
Duration: 07:16 (incremental)
Processing rate: 252 MB/s
03:34:57 Creating VM recovery checkpoint (mode: Veeam application-aware processing) 00:15
03:36:17 Hard disk 1 (60 GB) 2,5 GB read at 1 GB/s [CBT] 00:02

Now I see that, it can't be because of RCT. But where has the time gone?

Same VM (W2K22) Active full
Duration: 07:44
Processing rate: 544 MB/s
03:34:53 Creating VM recovery checkpoint (mode: Veeam application-aware processing) 00:14
03:36:11 Hard disk 1 (60 GB) 15,4 GB read at 544 MB/s [CBT] 00:29

OK, let's look again W2K22 (Domaincontroller) incremental

03:30:19 Queued for processing at 15.03.2023 03:30:19
03:30:21 Required backup infrastructure resources have been assigned
03:30:25 VM processing started at 15.03.2023 03:30:25
03:30:25 VM size: 60 GB (18,6 GB used)
03:30:25 VM is now in the required state for processing 00:00
03:30:26 Preparing to create snapshot 00:00
03:33:26 Using guest interaction proxy HOST2.xyz.local (Same subnet)
03:34:55 Inventorying guest system 00:01
03:34:57 Subscribing to guest processing components 00:00
03:34:57 Creating VM recovery checkpoint (mode: Veeam application-aware processing) 00:15
03:35:13 Using source proxy HOST1 (onhost)
03:35:56 Getting list of guest file system local users 00:00
03:36:01 Saving config.wmi 00:00
03:36:01 VMCX configuration file (50,2 KB) 50,2 KB read at 50 KB/s 00:00
03:36:06 VMRS configuration file (44 KB) 44 KB read at 44 KB/s 00:00
03:36:11 VMGS configuration file (4 MB) 4 MB read at 4 MB/s 00:00
03:36:17 Hard disk 1 (60 GB) 2,5 GB read at 1 GB/s [CBT] 00:02
03:36:24 Transaction logs will not be truncated 00:03
03:36:27 Saving GuestMembers.xml 00:00
03:36:30 Finalizing 00:00
03:37:18 Swap file blocks skipped: 1,6 GB
03:37:18 Deleted file blocks skipped: 232 MB
03:37:18 Busy: Source 74% > Proxy 71% > Network 65% > Target 52%
03:37:18 Primary bottleneck: Source
03:37:18 Network traffic verification detected no corrupted blocks
03:37:18 Processing finished at 15.03.2023 03:37:18

For comparison
Same VM W2K22 - VeeamZIP
Duration: 01.40
Processing rate: 386 MB/s
06:05:22 Queued for processing at 15.03.2023 06:05:22
06:05:23 Required backup infrastructure resources have been assigned
06:05:27 VM processing started at 15.03.2023 06:05:27
06:05:27 VM size: 60 GB (18,6 GB used)
06:05:28 VM is now in the required state for processing 00:00
06:05:28 Preparing to create snapshot
06:05:28 Creating VM recovery checkpoint (mode: Hyper-V child partition snapshot) 00:14
06:05:44 Using source proxy HOST1 (onhost)
06:05:48 Saving config.wmi 00:00
06:05:49 VMCX configuration file (49,4 KB) 49,4 KB read at 49 KB/s 00:00
06:05:54 VMRS configuration file (44 KB) 44 KB read at 44 KB/s 00:00
06:05:59 VMGS configuration file (4 MB) 4 MB read at 4 MB/s 00:00
06:06:04 Hard disk 1 (60 GB) 15,2 GB read at 386 MB/s 00:40
06:06:49 Finalizing 00:00
06:06:49 Backup file E:\VeeamZIP\VeeamZIP\dc1\dc1.vbk
06:06:50 Swap file blocks skipped: 1,6 GB
06:06:50 Deleted file blocks skipped: 541 MB
06:06:50 Busy: Source 90% > Proxy 80% > Network 59% > Target 97%
06:06:50 Primary bottleneck: Target
06:06:50 Network traffic verification detected no corrupted blocks
06:06:50 Processing finished at 15.03.2023 06:06:50

Does it have anything to do with the checkpoints?

Veeam uses production checkpoints as of W2K16. No matter what is set on the VM, right?

But VeeamZIP does not work when I disable the checkpoints and also VeeamZIP cannot do Application Aware Processing. VeeamZIP responds to the checkpoint settings on the VM.

Is it just like that and I don't have a bug?

Thanks and regards

[HannesK]

Hello,
hmm, looks all okay to me. As long as it's not causing an real problem, I would just leave it as it is. There are so many differences between Server 2012R2 and 2022, that it will be hard to find out the real reason why it takes longer.

Best regards,
Hannes