Anyone else doing Oracle & MS SQL log backups in same job?

[pkelly_sts]

[Veeam case ID 03252113]

Just curious if anyone else is backing up both Oracle & MS SQL DB servers in the same backup job?

Am currently working through a problem with support where for some reason the job seems to be not even attempting RPC comms to Oracle if we select log backups but goes straight to VIX which is then abysmal performance (kind of expected though I didn't expect it to be /this/ poor!).

One thing standing out to the Support guy is that this is the first time he's seen anyone backing up both in the same job. Not unsupported but unusual to him, so I thought I'd reach out here to see if anyone else is doing it?

Current suspicion (subject to a little more testing by me) is that it might be a new bug.

[Rick.Vanover]

Hi Paul - I've been working with SQL and Oracle as long as I've been at Veeam in some level and never did I think of doing it in the same job.

I'd be curious to know what the result is. I could think through some of the exclusions and credential management that could make it right, but usually I would realistically feel that having them different would be more "comfortable" - usually around RPO and storage considerations - yet I can also see how they could be the same, as they are both effectively production databases.

I have seen a job with SQL and Exchange in it - but that is more similar in terms of interaction.

Mind if I ask - why in the same job? (Seeking storage efficiency gains maximized, same RPO, there's only 1 job for everything, etc._)?

[Andreas Neufert]

We have this running in our SE demo lab and I am not aware of a limitation.

[pkelly_sts]

For us it's the simplicity of a single job per-site (3 sites) and likewise, I never thought to do it any differently based on technology

We're not a big org - two sites have 45-50 VMs being backed up, the 3rd only has 3 VMs.

SQL has its own retention/TL backup settings (we don't do TL backups for every DB server) and Oracle has its own. The fact that they happen to be the same doesn't really matter (essentially one backup/day & 15 min log backups is what we're looking for).

Whilst the Oracle system was being commissioned I had that in a separate job simply because it wasn't important enough at that stage to be copied off-site etc. (although i know you can pick & choose here too). I believe log backups worked fine in that configuration and I think it's only when we brought the Oracle environment into the main backup that we started experiencing issues (Oracle logs job not even attempting to use RPC & jumping straight into VIX once log backups are enabled)

Fundamentally I'm a big fan of Veeams "stick it all in one big job, configure the proxy/repository settings appropriately and let it just get on with it" approach compared to years of massaging BExec jobs trying to get timings right etc!

[Andreas Neufert]

If you had issues, then let´s fix those. What is the issue with the RPC processing not working and what is the error?
Did you open a support case and uploaded logs ?

[pkelly_sts]

Support case in the first post, it's being worked on by support but he hadn't seen the issue before (hence he suspects it's a bug) so I just thought I'd mention it here as it seemed to be an unusual (but not un-supported) use-case so I thought it might be useful to throw out there as an FYI if nothing else.

Basically without enabling log backups, RPC works fine. As soon as Log backups are enabled, RPC isn't even *attempted* for the Oracle VM but goes straight to VIX, which is then ridiculously slow and never completes, so the log-shipping server eventually fills with temporary logs (45Gb worth in my case).

[PTide]

As soon as Log backups are enabled, RPC isn't even *attempted* for the Oracle VM but goes straight to VIX.

This one is definitely not by design, - network connection has the first priority by default. I would assume that someone has changed the priority settings before, however if that was the case then the SQL VM would be also processed via VIX. Please keep working with the engineer and feel free to escalate the case if needed.

Thanks!

[pkelly_sts]

Yeah we changed the reg entry to force RPC but no joy.

Will report back here when we eventually sort it, I just thought I'd flag it up here in case anyone else was having similar issues.

[Andreas Neufert]

The 2 settings that I know are.

InverseVssProtocolOrder
Type: REG_DWORD (0 False, 1 True)
Value = 1 (enabled - VIX first)
Default value = 0 (disabled - \\admin$ first)
Desciption: needs to be added on the guest interaction proxy under HKEY_LOCAL_MACHINE\SOFTWARE\Veeam\Veeam Backup and Replication\


SqlBackupForceTransportMode
Type: REG_DWORD
Default value: 0
Path: HKLM\SOFTWARE\Veeam\Veeam Backup & Replication\
Description: Transport mode for SQL log backup transfer. 1 - direct transfer without shipping server, 2 - network transfer via shipping server, 3 - VIX transfer via shipping server

[pkelly_sts]

Sorry I thought I replied to this, mustn't have saved it.

We did try implementing the first reg entry with no difference (it's still currently in place) but as of now the second one doesn't currently exist (however as it says "...SQL log backup transfer..." does that imply it's related to MS SQL rather than Oracle?

[foggy]

Hi Paul, the second value works both for SQL and Oracle, so you can create it and see if it helps. Thanks.

[pkelly_sts]

OK, I'm assuming option 2 would be the first one to test? Will try today...