The backup copy job suddenly slows down on the last VM.

[Thomas609]

Hello everyone,

Our copy job transfers data to the offsite location through a dedicated 1 Gb/s network, and the storage device is StoreOnce. We have not configured any network traffic limits.

When performing the backup copy job, there are two VMs involved. The transfer speed for each VM is approximately 50 MiB/s. However, when the job reaches the last VM (which is not always the same VM), the speed suddenly drops to 10 MiB/s.

I couldn't find any relevant information about copy job speed in the documentation.

Could you please let us know if this behavior is normal?

Thanks

[david.domask]

Hi Thomas,

The total processing speed for a Backup Copy depends on the read speed of the source repository, network speed, and the write speed of the target repository.

What's the bottleneck listed as when the transfer speed drops down, and the overall bottleneck? I am doubtful it's network that's the bottleneck at first blush, and my guess is reading/writing is the bottleneck.

[Thomas609]

Hi David,

Thanks for your reply.

I reviewed the finished job details.
In the job summary, the bottleneck is shown as Network, but in the job details, the primary bottleneck is listed as Source.

This copy job includes twenty VMs. I checked the last VM that completed the copy process, and its throughput started at 40 MB/s, then dropped to 9 MB/s.

Also, I noticed that a backup job was running at the same time the copy job started. Could it be that the backup job is affecting the copy job’s throughput?

Appreciate your insights.

[david.domask]

Hi Thomas,

If the bottleneck is source, potentially yes the source job and backup copy running at the same time may impact the performance on the source repository for the Backup Copy. A fairly simple test would be to run the Backup Copy when the source job wasn't running and see if the overall processing rate increases. Keep in mind that the Concurrent Task limit will cap how many tasks can be processed at once, so if it's low for the source repository it may also be limiting the total throughput for the Backup Copy job.

[Thomas609]

Hi David,

I believe we’ve identified the issue. We have two gateway servers, and one of them is handling both backup jobs and copy jobs. Whenever the copy job runs on that particular server, its workload spikes to 99%.

We’re now looking into ways to resolve this bottleneck. Thank you very much for your suggestions and continued support.