I recently upgraded to V12 because, well, I had to. I put it off for awhile hoping all issues would be cleared by now. My jobs to my cloud provider now crawl. I do still have a V11 guest setup that does backups for one of our legacy apps. It was not pointed to my cloud provider previously. This morning I just pointed it and, yeah from the screenshots below it would appear that old issue in V12 has re-appeared. I do have a ticket open (TAC 07239898) but we got head scratching going on. TAC says its the network thats the bottleneck according to the logs. It seems like a false positive to me. The only factor that has changed is a version upgrade.
I'm not aware of any known issue.
Please get your cloud connect provider involved in the case. We will require logs from both sites (tenant + cloud connect provider) to have a full picture for further investigation.
One question, did you or the service provider used the "UseUnbufferedAccess" and "DisableHtAsyncIo" reg key in v11?
If so, the key was replaced by a single on in V12.* (DataMoverLegacyIOMode).
I don't see this key on my or the providers side. Should we try it? What does it do ? My V11 is a fresh install and doesn't have speed issues. Still having the issues with v12. Haven't got a solid explanation yet as to why there would be a difference. The PEX values between 11 and 12 are about the same. We can't figure out why there would be such a huge discrepancy in speed. I have a few 30T jobs I haven't started to send to the cloud yet because it will take forever at 100mbps.
Support still isn't offering any help with this issue. I pointed my veeam server to another cloud provider and its equally slow. I have no resort other then building a new veeam server. No idea how this update has hosed this machine. I'll report back if this was the solution
Hi set DataMoverLegacyIOMode to 1 ? I'll try that and see. I have not tried the manager thing yet. But I will if the registry key fails. Perhaps there is something to learn from this failed upgrade for Veeam.
Just an update, the Key didn't provide any improvement. We built a brand new Veeam 12 server migrated the configuration over sent a cloud job same dismal performance. As a test we created a brand new primary job to the local repo of the new Veeam 12 server and sent it to the cloud. It was a bit quicker but same dismal performance. Something up with veeamagent.exe in Veeam 12. In Windows Resource monitors my disks are basically idling as it reads data to send to the cloud. The devs are now going to take a look.
New update, we created a NTFS repo and performance to the cloud is acceptable, high speed. Moved the job to a ReFS repo, dismal performance. My guy is running with that along with dumps from veeamagent to ask how/why.
I got an update on my ticket. No forward progress, going a bit sideways actually. Im trying to explain that I got a test account on another cloud provider and Im able to duplicate the issue. I can't see now how they're saying its my primary cloud provider.
The investigation is still ongoing. Our QA team has requested all logs from the source (tenant backup server) and target (primary cloud connect provider). And the xperf logs. Please provide them with the requested logs.
First we want to confirm the cause of the issue and then we can also compare it to the other cloud connect provider facing the same problem.
+1 on interest in this issue. Wondering if it's related to the slow replication issue (see below). We are on 12.0.0.1420_20230412 and I was planning an upgrade to 12.1.2.172_20240515 but am hesitating because of the possibility of this issue.
OK. Sorry if I'm in the wrong space . I posted a reply on the other post and it was removed for being irrelevant to the topic. I saw this one and thought this was more precise to my question. I guess I just need to create a new topic .
I was using Backup Copy to a TrueNAS Core server connected with iSCSI. The server is a Supermicro with 8x 4TB SAS drives and a LSI SAS 12Gb Controller, but the hot swap backplane limited it to 6Gb. The ZFS was configured in a RAID 10 fashion for best write performace and used a Intel dual 10G ethernet card to a 10G switch where the Veeam B&R server was also homed. There are 9 Backup Copy jobs and the bottleneck was "Target". The performance after moving to v12 was slower. I tried upgrading the NAS memory from 24GB to 64GB, but no noticable difference.
I finally tried changing the NAS server in to a Linux Hardened Repository using a XFS repository and wow! I went from a processing speeds of 10-20 to 200-300 and new bottleneck became "Network".
My issue appears resolved, it said my bottleneck was "Network". When I upgraded from Veeam 11 to 12 "Use Multiple Upload Streams" was set to 4. Veeam 12 has changed how it sends data and for some reason 4 is not enough anymore to get full speed. Im at "20" streams and I now see full pipe bandwidth. I tried unchecking this option in V12 which made it worse, where in V11 it made no difference, I see full speed. Try changing this and see if it makes a difference for you.
kubimike wrote: ↑Jul 01, 2024 2:25 pm
When I upgraded from Veeam 11 to 12 "Use Multiple Upload Streams" was set to 4. Veeam 12 has changed how it sends data and for some reason 4 is not enough anymore to get full speed. Im at "20" streams and I now see full pipe bandwidth.
Try changing this and see if it makes a difference for you.
Thank you! We were seeing throughput of 80-100mbps with ours set to 5 so i've upped it to 20 and will monitor over the next few days!
