What the new feature offers is a simpler way to do what we've all been doing
anyway if we get throughput issues.

To make backup go faster you run many dsmc incr processes - one per
filesystem maybe. To make restore go faster you ensure that collocation by
filespace is on and run many dsmc restore processes.

This change allows simpler client configs and central scheduling - although
the scheduler could do with an overhaul.

Regards
Alan