We propose the possibility to render nodes inactive, without terminating them. This would basically mean that the server would just stop sending tasks to this node until it is activated again. Management and monitoring of the node would still be possible while it is inactive. This would enable additional flexibility from a toplogy point of view.

For example, let's say we have a computer with 4 CPUs and 4 nodes started on this machine. Depending on the kind of job that is executed on the grid, it would be very easy to switch from a topology with 1 node active with 4 processing threads, to one with 4 nodes active with 1 processing thread each, or any combination in-between. Dormant nodes could be reactivated very quickly, as this would only involve setting a flag on the server to let it send tasks to the activated nodes again.