Problem with management traffic SQL log backup

[dbr]

Dear Dev Team,

We back up multiple SQL server with SQL log backup at various remote sites. These sites are connected with a low-bandwidth wan link (such as 5Mbit) to the main site where the Veeam B&R Server is running. Actually, we don't need more bandwidth because the users in these sites mainly work with Citrix and for that purpose the link is sufficient. I opened a case (#02267597) because the SQL log backup uses the wan link to capacity every 15 minutes (approximately 40MB per backed up vm). Since this data isn't backup data between the data movers it isn't possible to be throttled via network throttling rules. Could you please consider to add a possibility to throttle management traffic as well or check whether the amount of data being sent in every 15-minutes-cycle for every backed up vm is really necessary or can be reduced to a minimum?

Daniel

[foggy]

Hi Daniel, my first guess is that log shipping server is being involved in the process. Can you confirm or deny that?

[dbr]

Hi foggy,

thanks for your reply. A direct connection is available between SQL vm and repository server and therefore no log shipping server is necessary (according to your help center link). Even if I select guest interaction proxy and log shipping server in remote site the amount of data being sent between proxy/repository server (a two in one vm in remote site) and Veeam B&R server remains the same. In one example the data is sent from veeam server port 50988 to proxy/repository server in remote site port 6160. Our network monitor tool shows 198MB of traffic between these ports every 15 minutes (default SQL log backup interval). According to help center port 6160 is use by Veeam installer service.

Remark: I'm on holiday the next three weeks. I will answer any replys once I'm back in office.

Daniel

[foggy]

Typically, there should not be such amount of traffic in the described configuration (where guest VM, repository and guest interaction proxy are in remote site, Veeam B&R in local site). Is it VMware or Hyper-V setup? Could you please check the size of transaction log backups metadata file in your repository (.VSM)?

[dbr]

We have a VMware only environment. The .vsm file has a size of 62MB.

[foggy]

I've checked the case and uploaded logs are not enough for investigation. Could you please collect the following:

- logs from Veeam B&R server: backup service logs, job logs and guest interaction proxy logs; server name/IP;
- logs from guest interaction proxy/repository server;
- logs from SQL Server VM, server name/IP.

You can ask your engineer for assistance in logs collection, upload the logs and ask for the case escalation.

Thanks.

[dbr]

Unfortunately, I had to disable SQL log backup for the site that was mentioned in the case because we cannot do quality of service at the moment. Fortunately we have other sites with SQL log backup, too. But they run wan connections with qos. Anyway the amount of data should be similar, so I collected log information from another site.

Did you notice the file capture_edited.txt / capture.pcapng on the old case ftp share? That was a capture of data within one 15 minutes sql log backup cycle. Please have a look at the files. In addition I uploaded a new capture and a prepared capture along with logs of the corresponding job to the new ftp folder. Additionally I informed my support engineer and he will escalate the case.

You can find servernames and ips in the case.

New case: 02326051

[foggy]

Thanks, I've also let our QA engineers know about the update.

[zercho05]

Hi foggy and dbr,

can you maybe let me know what the outcome of this case was? I know it's been quiet a while ago but still...
I have a similar problem with a big customer with about 150 remote Sites (with a Proxy and Repo Server on each Site) and and at least one SQL VM (with Log Backup) per site.

We are also experiencing a very strange behaviour with Log Backups.

The Log Backups of some sites are Freezing. and we were not able so far to find any clear hints on where the Problem is coming from.
Case #05587119

Thanks a lot
S.

[dbr]

It's a very long time ago but fortunately I have a documentation . The traffic consists of SQL log backup meta information (name of every single SQLOIBs stored in the corresponding SQL log backup of the backed up SQL server) and we kept SQL log backups as long as VM backup (30 days). That means every 15 minutes meta information of all SQL log backups is sent to the remote site. As a workaround we decided to decrease our RPO for our SQL log backup from 30 days (2880 SQLOIBs) to 2 days (96 SQLOIB). That's significantly reduce the amount of data sent over wan. I noted that Veeam is aware of this and will address this problem in one of the futures versions. But with VBR 9.5U3 the behavior was still the same. Meanwhile we have more bandwidth but also left the RPO at 2 days for SQL log backup. Never checked if that issue has ever been addressed in more current versions than 9.5U3. Hope this helps.