Feature request: task concurrency within jobs

[abognar]

It would be nice if the number of concurrent backups/replications could be limited within a job or concurrent processing of some VMs could be avoided.
For example if two domain controllers and are being processed at the same time (ending at the same time) then none of them will be available to server requests during snapshot removal.
This can be accomplished with either chained jobs or with a proxy where concurrent task number is one, but in the first case deduplication benefit will be lost, in the second case a proxy should maintained only for this purpose. Both need more management and care.

[Andreas Neufert]

Hi Job chaining would not be good as when one job stops the next one would not start.

Overall a DC should be compressed less than 15GB of backup data. I would create in this case another job at another time window (hours later) and backup this server. Anyway the trend goes to per VM chains because of the flexibility and so there is no deduplication between those servers. => Block cloning would still work to reduce data significantly.

Such a feature request would make much more sense for example to Exchange Server where you have great potential for deduplication when you backup 2 always on cluster nodes with the same DBs present.

[foggy]

To avoid concurrent backup of the two VMs in a single job, you could maybe play with the VM priorities in the job placing one of them at the top of the list and another at the bottom.

[abognar]

We have a custom orchestrator for running multiple jobs which can also chain them where a job failure doesn't cause any problem but we will probably deploy a Linux proxy for this purpose. I don't like timing for this purpose as multiple tasks are depending on the backups (for example replications which could also cause such problems).

Yes, this feature request makes more sense from a storage perspective in case of an exchange server or anything else with more data then a domain controller, but we encountered this problem in case of two domain controllers. It would be much simpler to change a value within the job settings than to handle this situation with job management