DAGwoman - DAG-based WOrkflow MANager

Introduction:

DAGwoman is a workflow engine that allows running and managing scientific workflows on computing resources.

DAGwoman aims at enabling the application of DAG file based scientific workflow in computing environments where neither Condor or Condor-G nor GridWay is installed and, thus, not available. DAGwoman is implemented in Java and can be run in user-space, that means, there are no administrator permissions needed to install the software.

Workflows have to reflect directed acyclic graphs and are specified in the DAG file format as defined for the Condor/DAGMAN engine, also used by the GridWay/GWDAG engine.

Currently there exist two interfaces which can be used with DAGwoman to interact with the installed scheduling system or middleware. If your compute resource runs Sun Grid Engine (SGE) or nowadays Oracle Grid Engine (OGE) their native commandline interface can be used, i.e., DAGwoman uses qsub, qstat and qacct to submit to and query SGE/OGE. More performance you usually get if your compute resource runs a scheduling system or middleware that implements Distributed Resource Management Application API (DRMAA) with Java bindings. In this case DAGwoman can directly interact with the scheduling system.

Reference:

DAGwoman has been presented at the SWEET2012 workshop of the SIGMOD conference in Scottsdale (AZ, USA) and was published as

• T. Tschager and H.A. Schmidt (2012) DAGwoman: enabling DAGman-like workflows on non-Condor platforms. Proceedings of the 1st ACM SIGMOD Workshop on Scalable Workflow Execution Engines and Technologies (SWEET 2012), article 3. (DOI: 10.1145/2443416.2443419)

Download DAGwoman:

• the DAGwoman software is available upon request to heiko.schmidt(AT)univie.ac.at