Workflow

Tony Linde ael at star.le.ac.uk
Wed Jan 19 14:49:12 PST 2005


If we really need to access distributed data on a data grid then we should
extend our data access standards to allow access to grid-based data.

We don't need our *own* standards for grid-based computation, just standards
for describing such services and a way of invoking them that is an extension
of normal service description (which we need more urgently).

And if a job is submitted from a portal, I expect that portal/workflow
system will have its own way of interacting with the executing job. There is
no real need at this stage for that to be standardised. 

None of this argues for workflow standards, rather it argues against them as
it highlights the critical areas where we really do need standards now.

Cheers,
Tony. 

> -----Original Message-----
> From: owner-interop at eso.org [mailto:owner-interop at eso.org] On 
> Behalf Of Reagan Moore
> Sent: 19 January 2005 19:35
> To: Interop IVOA
> Subject: Re: Workflow
> 
> Tony:
> The reason for interoperability between workflow environments 
> is contained within your message.
> 
> >
> >Any project which wishes to develop a workflow creation, 
> submission & 
> >execution tool would write the appropriate software so that a user 
> >could select from a set of tools and data sources (from the 
> registry) 
> >and string them together with some flow logic into the 
> workflow. This 
> >would then be submitted to a job execution service etc.
> >
> 
> 
> A reason for workflow interoperability is to minimize the 
> amount of software that must be written to construct a new 
> data processing 
> pipeline.    Workflow systems interact with:
> - data grids for managing distributed collections
> - grids to support distributed computation
> - portals to organize interactive access to the combined systems
> 
> Examples of workflow systems that simplify processing (and 
> interact with the above environments) include:
> - Kepler/Ptolemy - this provides actors for manipulating 
> iterative processing.  The computation can be done locally or 
> remotely.
> - Matrix - provides a dataflow language that simplifies 
> description and management of state information within the 
> dataflow system
> - Chimera/Pegasus - supports mapping of workflow environments 
> to grid technology for distributed computation
> - Condor DAGman - supports processing of directed acyclic graphs
> 
> These systems are developing interoperability mechanisms for 
> interchanging workflows.
> 
> Reagan Moore
> 



More information about the interop mailing list