Modeling grid workflows with colored Petri nets

Abstract. Grid computing refers to the deployment of a widely distributed architecture for the execution of computationally challenging tasks. The grid provides a set of distributed resources which can be used for "computing on demand " or for constructing a "virtual supercomputer". Recently, several researchers started to look at the relation between workflow management and grid computing. The flow of work through a grid can be seen as a classical "workflow". However, as opposed to the classical workflows, the resources are not humans and are not managed by some centralized client-server architecture. Instead, the grid is highly distributed and the resources are computing power, memory, etc. Currently, there is no conceptual framework for grid computing and the role of workflows in grids is unclear. This paper provides initial steps towards a conceptual framework expressed in terms of Colored Petri Nets. CPN Tools is used to model grids while focusing on the workflow aspects. The resulting model can be analyzed to detect deadlocks, etc. The framework is illustrated using process mining as an application.