Abstract
This chapter shows how flexibility can be realized for distributed workflows. The capability to dynamically adapt workflow instances during runtime (e.g., to add, delete or move activities) constitutes a fundamental challenge for any workflow management system (WfMS). While there has been significant research on ad-hoc workflow changes and on related correctness issues, there exists only little work on how to provide respective runtime flexibility in an enterprise-wide context as well. Here, scalability at the presence of high loads constitutes an essential requirement, often necessitating distributed (i.e., piecewise) control of a workflow instance by different workflow servers, which should be as independent from each other as possible. This chapter presents advanced concepts and techniques for enabling adhoc workflow changes in a distributed WfMS as well. Our focus is on minimizing the communication costs among workflow servers, while ensuring a correct execution behavior as well as correctness of ad-hoc workflow changes at any time.
TopIntroduction
For a variety of reasons enterprises are developing a growing interest in aligning their information systems such that they become process-aware (Lenz, 2007; Müller, 2006; Mutschler 2006; Mutschler, 2008a). Such process-aware information systems (PAISs) offer the right tasks at the right point in time to the right actors along with the information, resources and application services needed to perform these tasks (Dadam, 2000). Business process management technology offers promising perspectives to achieve this goal (Weske, 2007). Examples include workflow management systems and case handling tools (Günther, 2008 a; Mutschler, 2008b).
A workflow management system (WfMS) enables computer-supported business processes (i.e., workflows) to be executed in a distributed system environment (Bauer, 1999; Muth, 1998; Shegalov, 2001). Usually, a WfMS provides powerful tools for implementing enterprise-wide, process-aware information systems (PAISs) (Dadam, 1999). As opposed to data- or function-centered information systems, a WfMS separates the specification of the process logic (i.e., the control and data flow between the process activities) from application coding (Dadam, 2000; Weber, 2007); i.e., process logic can be described explicitly in terms of a workflow template providing the schema for workflow enactment (workflow schema for short). The different activities, in turn, are implemented as loosely coupled application services that can expect that their input parameters are provided upon invocation by the WfMS and which only have to produce correct values for their output parameters. Usually, the core of the workflow layer is built by the WfMS which provides generic functions for modeling, configuring, executing, and monitoring workflows.
Key Terms in this Chapter
Compliance Criterion: refers to a well-established correctness criterion for adaptive workflows that can be applied to check whether a running workflow instance is compliant with a changed workflow schema; i.e., whether the change is valid for the workflow instance. For example, compliance will be always ensured if the execution log of the respective workflow instance can be produced on the new schema as well.
Migration: refers to the transfer of the control over a particular workflow instance from one workflow server to another.
Workflow Change History: refers to a log which captures all change events related to a particular workflow instance. Thus it complements the execution history of the instance which logs start and completion events of activities.
Workflow Change Pattern: allows for workflow schema adaptations at a high level of abstraction. Examples include high-level changes like the insertion, deletion and movement of process schema fragments. Workflow change patterns can be also used to assess the expressiveness of a change framework.
Distributed Workflow: refers to a workflow whose schema is subdivided into several partitions which are then controlled piecewise by different workflow servers. Accordingly, a distributed workflow management system (WfMS) is made up of several workflow servers that allow for such distributed workflow execution.
Adaptive Workflow: refers to the ability of the workflow management system to dynamically adapt the schema (i.e., model) of in-progress workflow instances during runtime.
ADEPT: refers to an adaptive process management system developed at the University of Ulm. The ADEPT technology allows for dynamic workflow changes at different levels. It enables ad-hoc changes of single workflow instances as well as workflow type changes and their propagation to running workflow instances.
Ad-Hoc Workflow Change: refers to a change applied in an ad-hoc manner to a workflow instance (e.g., by dynamically adding, deleting or moving activities). Usually, ad-hoc changes of a workflow instance become necessary to deal with exceptions or unplanned situations not anticipated at workflow design time.