Network crawling SOBR Synology, Backblaze B2

[deepiceman]

Hi,

I am using Veeam to backup a physical server onto a Synology NAS using the SMB unc file share at the moment. It is configured using the instructions to offload to a Backblaze B2 SOBR with immutability.

Synology has a similar article with Object storage.

When the backup runs it is slow but does finish in 4-5 hours (about 3TB in size). However, when the SOBR tiering starts, it renders my network useless. Nothing on the network works and the processing rate on the job is maxed out at 5 MB/s - yes MB/s. The bottleneck is 99% at the target.

The steps in the article require creating a local Veeam repository (local hard disk), followed by an S3-compatible repository, and finally, the scale-out repository connecting them. In my case, my local Veeam repository is the SMB unc share on Synology. I am certain that is the bottleneck here and causing the network issues.

How do I fix this? I know iSCSI is recommended over SMB and I am willing to go that route but will that work or cause the same issue? Can I use the proxy option to make it faster?

Any ideas to remedy this is appreciated. Thank you.

[Mildur]

Hello Mandeep

How do I fix this?

Unfortunately, these cannot be investigated through a forum post. Technical and performance issues should be addressed by our customer support team, as outlined in our forum policy. Please open a support case and share the case number with us. Thank you for your understanding.

When the backup runs it is slow but does finish in 4-5 hours (about 3TB in size)

Without knowing your job configuration or server/network design, one potential solution could be to use iSCSI connected to a Windows or Linux repository server. SMB repositories often fall short in performance, especially during synthetic full or forward incremental operations. This is because all the backup data blocks from entire full backups must be copied and rewritten between the gateway server and the SMB repository, leading to longer session durations.

However, when the SOBR tiering starts, it renders my network useless.

You can manage network consumption by applying throttling policies. Alternatively, you can configure a specific time window during which the Offload to Backblaze B2 is allowed, making sure it runs at times that are less disruptive to other network activities.
- Internet Throttling policy
- Offload Window Setting

Best,
Fabian

[deepiceman]

Thanks for the reply Fabian. This is my second install and it is doing the exact same behavior as my first one. Last year I opened a support case (#06292151) for it and nothing came out of it. In that case, the backup of nearly 3 TB finished after we let it run through its course and since then the following updates have always gone through without any issues as the uploads are so small. We do full backups weekly but it never resends the full 3 TB data over again.

My intention to open this on the forum was to see if there was another soul out there who had the same issues and figured it out. My network is very simple - ISP - Router/Firewall - Switch - Devices. I have switched from SMB to NFS and there isn't much difference in outcomes. I did a packet capture while the SOBR was running and it is littered with retransmit and duplicate acknowledgement packets.

Next items to try on my hit list are network throttling, I/O throttling, and maybe separating the NAS and Veeam on a separate VLAN.

Thanks