Automatic workflow Generation for GridSAM

Google Summer of Code 2008 ideas

Primary Mentor: Stephen McGough (asm at doc.ic.ac.uk)
Secondary Mentor: Vesso Novov
OMII Project: GridSAM.

Description

The GridSAM project is an open source project for making resources available through a standardised job submission interface BES using a standardised job submission language JSDL available from the Open Grid Forum (http://www.ogf.org). GridSAM provides a veneer over existing resources (either clusters or single computers) which translates the incoming standardised job submission into the format required by the underlying architecture. Many users before they start using remote computational clusters (the Grid) will have performed job execution on their own local resources. Inevitably they will have developed scripts for executing more than one task. In general when submitting jobs into such a Grid environment the user could submit their script where all the tasks will run together on the same computational resource. However, this looses a lot of the potential from the Grid where many resources are made available to the end user. In the ideal scenario the user would break their original script down into separate tasks and use a workflow language to submit this to the Grid. Unfortunately this tends not to happen as the users are too busy to do this. The aim here is to provide tooling which can take the users script and automatically decompose it into separate tasks which can be submitted separately through GridSAM. Effectively translating their script into a workflow. This would allow independent tasks within the script to be executed simultaneously.

Project Requirements

A competent programmer in Java who is experienced in at least one of Perl or a Unix shell environment. Experience in using XML and Java programming of Web Services would be desirable.