Various embodiments enable a server system to consider a variety of
priorities in determining how to organize, allocate, and manage client
workloads in a distributed processing system. In at least one embodiment,
a method may include generating a capability vector for one or more
distributed devices, the capability vector defining a capability of the
one or more distributed devices, determining a client priority for one or
more client system, the client priority defining an order that a client
system is granted access to submit one or more workloads to a distributed
processing system for processing, assigning a workload priority to one or
more workloads, the workload priority defining an order in which a
workload is submitted to the distributed processing system for
processing, and distributing one or more workloads form a client system
to one or more distributed devices in response to the capability vector,
the client priority, and the workload priority.