[Beowulf] WORKS 2006 Call for Participation

Douglas L Thain dthain at cse.nd.edu
Tue Jun 13 09:12:48 PDT 2006


Panel on "Workflow as the Methodology of Science"
Tuesday June 20 2006 WORKS Workshop 
HPDC Paris France 12pm - 1.30pm
http://www.isi.edu/works06    
Moderator Geoffrey Fox

A recent NSF workshop http://vtcpc.isi.edu/wiki/index.php/Main_Page
proposed that workflow could be viewed as  underlying support for the
scientific methodology emerging in many fields and involving distributed
interdisciplinary data deluged scientific methodology as an end
(instrument, conjecture) to end (publication, archived results) process.
This vision for workflow mixes the coupled execution of related services
characteristic of most scientific workflow with the more asynchronous
longer term processes familiar in some business workflow. Can one
usefully link these different styles of workflow and further enhance
scientific productivity?

One challenge is reproducibility of this full process which is core to
the scientific method and requires rich provenance, interoperable
persistent repositories with linkage of open data and publication as
well as distributed simulations, data analysis and new algorithms. The
distributed reproducible science methodology can be supported by
publishing all steps in a sort of electronic logbook that is the
"script" of the full scientific workflow. It would need to capture the
scientific process (data analysis) as a rich cloud of resources
including emails, presentations, wikis as well as databases, compiler
options, build time/runtime configurations etc. One could need to
separate wheat from chaff in this electronic record (logbook) keeping
only that required to make process reproducible and allowing selective
execution (checking) of components of the log.

Is this a reasonable picture for a future workflow requirement and what
are the new research challenges it engenders?

The presentations at NSF meeting can be found at
http://vtcpc.isi.edu/wiki/index.php/Documents and give us a starting
point!

Contributors:
E. Deelman, USC/ISI, Summary of NSF Workshop
S. Jha,	University College of London, Application perspective
D. De Roure, University of Southampton, Provenance
I. Foster, ANL & Univ. of Chicago, Lessons from current Science Grids
John Ibbotson, IBM Hursley, Web service and business workflow



More information about the Beowulf mailing list