Re: [wsbpel] Q: process priority?

[Ron Ten-Hove <Ronald.Ten-Hove@Sun.COM>]

Sid,

    Your question brings to my mind another question (isn't that how scientists stay in business? :-) ). What kind of priority to you mean? There are two meanings that I can think of in our BPEL context:

• Business priority. The equivalent of a rush order, meaning that preference should be given to the order ahead of "normal" priority orders.
• Technical resource priority. Preferentially assigning computing resources to a particular process, or, alternatively, assigning only otherwise idle resources to a "nice" low-priority process.
  

Of course there could be overlap; we me choose to give greater technical priority to a high business-priority process, as a means of  fulfilling the business goal.

    Both concepts have a place in the world, but I suspect the "technical" priority is more of an implementation issue. Business priority often shows up in workflow management systems, but BPEL's sparse model of work items makes it difficult to discuss relative business priorities; there is no real place where work items queue up, and can be prioritized. In other words, I don't see how business priority can be reflected in a BPEL model of a process at run-time. Perhaps that's just my lack of imagination... :-)

    But I digress. What did you have in mind when you used the phrase "process priority?"

Cheers,
-Ron

Sid Askary wrote:

Since we are dealing with an execution language, the question of runtime priority comes to mind.  One could imagine an additional process property.  While there may be other methods, this could explicitly be expressed as an additional binary attribute to the process element followed, conditionally, by a new element.

a)
<process  name="ncname" targetNamespace="uri"
    queryLanguage="anyURI"?
    .
    .
    prioritize="yes|no"?/>

b)
<priority name="ncname"
    priorityNumber="anynumber"
    PriorityNumberBase="anynumber"/>

NOTE:  PriorityNumberBase reflects the devisor for PriorityNumber.  Alternatively, one could replace the two by a single floating-point fraction.


What does the group think?

Thanks,
Sid.



To unsubscribe from this mailing list (and be removed from the roster of the OASIS TC), go to http://www.oasis-open.org/apps/org/workgroup/wsbpel/members/leave_workgroup.php.