Setting job precedence/stagger times

  • 21 November 2022
  • 4 replies

Userlevel 4
Badge +12

Good day all

I guess this question is two-fold and I’ve not really been able to find a way this can be done using a Plan.

We’re running SQL Plans that backup the DB’s in the traditional Full/Diff/TLOG format.

We have 1 SQL instance that mustn’t have it’s TLOGS backed up. My assumption is that it would need to go into it’s own Plan with TLOG slider switched off?

The second portion to this same instance is that the backups of the various DB’s don’t happen at the same time due to their size?
Is it possible to configure an instance that a DB will not start backing up until the prior one is complete and they keep cascading in this manner.
I understand you could create multiple sub-clients within each instance and assign different schedules. However, this would potentially mean that if a DB finishes sooner than the next one is scheduled to start, there’s an unused portion of the backup window. Conversely if one schedule overruns into the next, then we have 2 jobs running which we’re trying to avoid.
The only way I can think of this is to limit the stream count to 1 on an instance, and the DB’s will then queue up. This would impact performance though, too.




Best answer by Luca Marinoni 22 November 2022, 12:02

View original

4 replies

Userlevel 6
Badge +15

Good morning.  the default behavior would be that the job runs like a normal subclient so 1 database would run at a time.  Is this an issue that this is multiple clients and you would like one to run at a time?

Badge +3

Hi MAuro,

For the first point I usually create 2 subclients one for simple recovery model DB backup and the other for Full using so the different schedules one with TLs and the other without.

For point two, if you want to prioritize the DB backups you could make a subclient for each DB and then use a workflow to start them as soon as the previous ones finish, but it is not that simple.

Userlevel 4
Badge +12

Thanks all for the feedback. Will be testing workflows for this scenario.

Userlevel 7
Badge +19

What a workarounds for such a request! I honestly also do not understand why its not there and also miss this ability to add a randomizer/staggering in plan configuration. Do not that if you change the RPO to hours than it will overtime flatten out the start time but that takes a long time a manual scheduling of jobs to influence the scheduling. We still see a lots of job being fired-off at the same time like "traditional” time-based scheduling. Just remembered I didn't open a CMR for it, so I'll do that tomorrow.