Say you have 3 drives.
Target sessions for each is 4.
Max sessions for each is 16.

Now we have incoming sessions.
First 4 sessions go for example to drive 1.
Next 4 sessions go for example to drive 2.
Next 4 sessions go for example to drive 3.

What happens to the next session?
And the one after that?
And the one after that?

https://www.dell.com/support/manuals/en-us/networker/nw_p_nwadmin/target-sessions?guid=guid-affc15ce-2b22-434a-8995-1e25ebf685d3&lang=en-us gives the following information:

When a save session starts, the following actions occur:
* If a device is already receiving the number of backup sessions determined by the target sessions value, the NetWorker server uses the next underutilized device for the backups.
* If all available devices are receiving the number of backup sessions determined by their target sessions value, the NetWorker server overrides the set value and uses the device with the least activity for the next backup session.

Now especially concerning the second part, in reality it seems to work like follows (which IMHO is in contradiction with "device with the least activity"):
Next 12 sessions go to drive 1.
Next 12 sessions go to drive 2.
Next 12 sessions go to drive 3.

My problem with this is that this will never maximize throughput.
How is NW supposed to know how many more sessions are inbound if the goal would be to spread the load across all available drives? (Without a crystal ball, it can't.) It would make more sense if it did:

Next session goes to drive 1.
Next session goes to drive 2.
Next session goes to drive 3.
Next session goes to drive 1.
Next session goes to drive 2.
Next session goes to drive 3.
etc.

But of course it depends.

There are slow savesets that can just be interleaved.

But with large and fast savesets it would make sense to use as many drives as possible right away.

Thoughts? (Except migrating to backup-to-disk)