Continuous Workflows: From Model To Enactment System
Panayiotis (Panickos) Neophytou
Monday December 17, 2012
10:30 am - 5317 Sennott Square
ABSTRACT
Workflows are actively being used in both business and scientific domains to automate processes and facilitate collaboration. A workflow management (or enactment) system (WfMS) defines, creates and manages the execution of workflows on one or more workflow engines, which are able to interpret workflow definitions, allocate resources, interact with workflow participants and, where required, invoke the needed tools (e.g., databases, job schedulers, etc.) and applications. Traditional workflow enactment systems and workflow design processes view the workflow as a one-time interaction with the various data sources, i.e., when a workflow is invoked, its steps are executed once and sequentially. The fundamental underlying assumption has been that data sources are passive and all interactions are structured along the request/reply (query) model. Hence, traditional WfMS cannot effectively support business or scientific monitoring applications that require the processing of data streams such as those generated nowadays by sensing devices as well as mobile and web applications.
It is the hypothesis of this dissertation that Workflow Management Systems both in the scientific and business domains can be extended to support data stream semantics to enable monitoring applications. This includes the ability to apply flexible bounds on unbounded data streams and the ability to facilitate on-the-fly processing of bounded bundles of data (window semantics). To support this hypothesis this dissertation has produced new specifications, a design, an implementation and a thorough evaluation of a novel Continuous Workflow model, which is backwards compatible with currently available workflow models. We implemented our proposed Continuous Workflow model, in a system called CONFLuEnCE, as an extension of Kepler, which is a popular workflow management system widely used in the scientific domain. In order to support the requirements of different applications, we developed a scheduling framework for implementing different policies for the execution of workflow steps. We demonstrated the applicability of our model in both scientific and business applications by utilizing CONFLuEnCE in Astroshelf to support live annotations (i.e., monitoring of astronomical data), and to support supply chain monitoring and management, respectively. Finally, we evaluated the performance of our scheduling framework using the Linear Road Benchmark for continuous workflows.
DISSERTATION ADVISER
Dr. Dr. Alexandros Labrinidis, Dr. Panos Chrysanthis, Department of Computer Science
COMMITTEE MEMBERS
Dr. Kirk Pruhs, Department of Computer Science
Dr. Calton Pu, College of Computing, Georgia Tech