FW: Workflow

[Tony Linde]

 Jim's reply...

-----Original Message-----
From: Jim Gray [mailto:gray at microsoft.com] 
Sent: 19 January 2005 18:33
To: Tony Linde
Subject: RE: Workflow

I have been saying for 35 years that we need workflow standards. 
Actually, about 20 years ago I realized the folly of it -- getting people to
agree is hard if there is no incentive to agree. 
What has happened is that two common workflows are sort of standardized:
  Request-reply
  Transaction
  Enqueue-Dequeue
I think WS-Orchestration is a nice start on something that many people could
agree to. 
But, it really depends on what people want to do. 
Pipelines have one set of needs, Condor has another, eMail has a third,
BizTalk has its goals,  and so it goes. 
You are old enough to remember batch job schedulers (perhaps even JCL). 
We never managed to standardize that  --- actually there were many standards
but we all ignored them. 
So... I am with you .

Jim Gray
Microsoft Research,  Suite 1690, 455 Market, SF CA 94105,
tel: 415 778 8222 fax: 425 706 7329
Gray at Microsoft.com   http://research.Microsoft.com/~gray

-----Original Message-----
From: owner-interop at eso.org [mailto:owner-interop at eso.org] On Behalf Of Tony
Linde
Sent: Wednesday, January 19, 2005 9:50 AM
To: Interop IVOA
Subject: Workflow

(I'm assuming Interop is the right list for this message.)

It was mentioned during the exec meeting/telecon that IVOA ought to look at
standards for workflow. I must say that I think this would be an unnecessary
drain on our resources when we already have too many things to work on
standardizing.

Presumably any such standard would state the way that a workflow ought to be
described so that it could be submitted to some unnamed workflow engine for
execution. But why would we need such a standard?

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.

The only reason for standards in the workflow arena is if we expect that
people willl want to create a workflow using one project's tools and then
submit it to the job execution service run by another project. I think this
is highly unlikely and certainly not something that will gain us sufficient
benefits that we need to push effort into it now.

We already have a wide range of efforts proceeding: we need more registry
standards, more data models, more data accecss standards as well as the new
events effort. I really think that to start, or even start discussing,
workflow standards at this point is superfluous.

Can someone persuade me that we do need workflow standards?

Cheers,
Tony. 

__
Tony Linde
Phone:  +44 (0)116 223 1292    Mobile: +44 (0)7753 603356
Fax:    +44 (0)116 252 3311    Email:  ael at star.le.ac.uk
Post:   Department of Physics & Astronomy,
        University of Leicester
        Leicester, UK   LE1 7RH
Web:    http://www.star.le.ac.uk/~ael
            
Project Manager, EuroVO VOTech   http://eurovotech.org 
Programme Manager, AstroGrid     http://www.astrogrid.org 
Co-Director,
 Leicester e-Science Centre      http://www.e-science.le.ac.uk/