Knowledge-based Worfklow System for Grid Applications
Project Co-ordination: Steffen Unger,
Fraunhofer FIRST
http://www.kwfgrid.net/
Research
The main objective of the K-Wf Grid project is to enable the knowledge-based support of workflow construction and execution in a Grid computing environment. In order to achieve this objective the consortium members will develop a system that will enable users to semi-automatically compose a workflow of
Grid services, execute the composed workflow application in a Grid computing environment, monitor the performance of the Grid infrastructure and the Grid applications, analyze the resulting monitoring information, capture the knowledge that is contained in the information by means of intelligent agents, and finally to reuse the joined knowledge gathered from all participating users in a collaborative way in order to efficiently construct workflows for new Grid applications.
The main research topics in the domain of workflows are:
- Innovative workflow concepts for enabling dynamic and interactive Grid workflows
- Grid workflow description languages based on Petri nets and XML, supporting several levels of abstraction
- Grid workflow orchestration and execution environment that allow users to assemble and invoke workflows on dynamic Grid resources, concerning interactive alterations and refinement of the workflow during runtime
- Semi-automatic mapping of abstract workflows onto concrete Grid services with the assistance of an expert system
- Advanced monitoring and performance analysis of workflows
Approaches
The K-Wf Grid consortium proposes the following approaches in order to meet its specific needs:
- The semantic description of resources will mainly be based on OWL-S and stored in a distributed knowledge base called Grid Organizational Memory (GOM) which will be optimized for RDF data. We will use RDQL as a query language for ontological data.
- The Grid workflow description language (GWorkflowDL) will be based on coloured Petri nets and XML.
- The workflow enactment machine will allow automatic or interactive refinement of the workflow during runtime by Petri net refinement.
- The expert system and the Grid Organizational Memory will use JESS and/or JENA (Java Semantic Web Toolkit) and OWL-DL for the description logic.
- The monitoring will enable the dynamic instrumentation of Java-based Web Services and C/Fortran legacy code.
- Most of the system components will be implemented in Java and deployed as a Web or Grid Service.
- To be WSRF-compliant we will make use of the Globus Toolkit 4 with the option to later switch to gLite.
- The user interface will be based on JSR-168 portlets, OGCE (GridSphere), and Globus components.
Related Technologies
BPMN
b_bprimal