Whenever a designer requests that a part of a design be brought up-to-date, (such operation being known as retracing), VOV schedules all necessary jobs and dispatches them to the available slaves.
At the same time, other designers may request the retracing of other parts of the design. These multiple requests may in part or in whole overlap with existing requests, and VOV guarantees that no job will be executed more than once.
Each slave is characterized by the resources it offers, by the relative power of the CPU, and by the load on its host, which can be updated every minute. The dispatching rule is that the longest job goes to the most powerful slave offering all required resources. The resources may include, for example, software licenses, amount of physical memory, access to special devices like printers and accelerators.
The network may be heterogeneous in many ways. First, there may be machines from different vendors. VOV supports the equivalent of a parallel "make" for multiple architectures. A second source of heterogeneity comes from the many possible ways in which file systems may be mounted across the network. It is possible that the same file may have different names on different machines. VOV understand this and maps all names to a "logical canonical" name that is valid across the network.