UNthrottling mid-job?

[ChrisL]

I've kicked off a large backup job at the same time as another job. The Repository (QNAP NAS over iSCSI) has throttling enabled on it to stop it from crashing and killing everything off in the process..!

The large job has one VM, 4HDDs. The other job is smaller in total size but has about 11VMs with at least 2HDDs each. I'm guessing that at the start, the throttling rules kicked in and limited the transfer rate to keep the QNAP happy (good thing!) but now that the smaller job has finished, the larger job appears to still being subject to the original throttling, despite there now being plenty of bandwidth avaialble.

Does Veeam not monitor the throttling throughout the job and try to maximise it for the given bandwidth at that time. At the moment I have a large Active Full job running at only about 30MB/s with the Task Manager Network graphs showing traffic at about 25% where it could be going a whole lot faster.

[veremin]

What value have you set for network throttling rules? So far it sounds like an expected behaviour - you've set throttling rules and jobs respect those (meaning they limit processing speed to a certain value). Thanks.

[ChrisL]

I had it set to 80MB/s on the repository. At the start of the process, when both jobs were running, it was probably hitting this, and so the throttling was doing its job correctly by balancing the available bandwidth between the running jobs, each getting about 30-35MB/ to make best use of it.

My question is what happens when one of the jobs finishes, and what happens to the job that's still running. It appears to be still subject to the initial throttling, stuck at 30-35MB/s. But since it's the only job now running, it should be able to speed up to use the full 80MB/s..?

[foggy]

Hi Chris, are you referring to network traffic throttling or repository read/write data rates?

[ChrisL]

Hi Foggy, the second one, the settings on the repository itself, not on the proxy

[foggy]

This throttling mechanism is able to dynamically re-balance the load. Most likely the reason for the observed behavior is that some other component is the bottleneck for this job, which doesn't allow it to process data faster. You can try to clear the checkbox in repository settings and run a single job to see what is reported as the primary bottleneck for it.

[ChrisL]

Thanks Foggy, yeah I wondered if that might be the case.

The job details for the big job though indicate the bottleneck as being 'Throttling'
Source - 49%
Proxy - 40%
Network - 69%
Target - 61%

I do have encryption enabled on the job which I didn't have before. I wonder if this is impacting the performance at all. The Performance of the host computer isn't particularly overloaded though. CPU running between 20-30% usage over 16 cores and RAM steady at 14GB out of 16GB physical, so nothing really saturated there either.

[foggy]

I'd check bottleneck stats for a single running job (without any other parallel tasks) with throttling disabled. Also, what server is acting as a proxy in this case? The default one (backup server)?