BPMN and BPSS

[Stephen A White <wstephe@us.ibm.com>]

Here are some initial comments and questions
on the example that was presented to the BPMI Notation Working Group. We
may have more comments and questions as our group continues discussions.
A new version of the example is attached to illustrate some of the comments.

-Steve

·        A
modeler might want a model of just the center section of the diagram by
itself as a “global” process, which can be modeled as is by BPMN, except
for the diamond (see next). Issues with the example are:
o        The
diamond (Decision) in the middle would be what we call an Event-Based Gateway—there
is already a specific representation for this type of Gateway.
o        The
Flow that shows the PO and the ACK docs should not be shown as Sequence
Flow (solid lines). It should be Message Flow (dashed lines). The Document
Objects could then have an Association with the Message Flow.
o        Transaction
(double line) activities: This fits with current BPMN, although it is only
for a Sub-Process. Including it for a Task should not be a big deal.
·        A
modeler might want a model of either the left or right sections of the
model by themselves as being an abstract process (or even a complete process).
These can be modeled by BPMN, except for the endpoint objects (see below).
Issues in the example are:
o        Each
of the Pools, on the left and right sides, should have its complete Sequence
Flow. It is understood that this is an example and not necessarily complete.
o        The
clouds could be BPMN Artifacts, but should not be a part of the Sequence
Flow. They should be attached to the flow through Association lines. 
o        Endpoints
(small circles on boundary of Pool) have to be considered more carefully.
Are they intended to organize/correlate sets of messages that are part
of a specific exchange? If so, BPMN does not have a graphical object for
this (although Associations could be used to convey this information—but
not in a compact way). This could be an issue for extension.
§        Looks
like a Start Event—thus, the shape should be changed. This new shape would
have to be added to BPMN and its properties and semantics defined. Perhaps
a hexagon shape.
§        Alternatively,
they could be a type of Intermediate Events that represent the sending
or receiving of a message, except that the model shows them connected to
the Tasks that send and receive the message. Also, an Intermediate Event
is only one way and could not be an invoke. (This could be an issue for
extension).
§        They
probably should be connected to the source activities through an Association.
·        A
modeler might want a model with both the left and right sides together
and the Message Flow between them. BPMN can model this now without the
endpoint objects. This would be sort of a distributed “global” process.
It would lack the relationships between the messages to see how they could
be organized together—presumably what the endpoint objects are for.
·        A
modeler might want to combine all of these elements into one model, as
in the example. Since BPMN allows a modeler to define the meaning of a
Pool, the current example can be modeled almost completely with BPMN as
it is now. Issues with the example are (some have been described above):
o        Proper
use of the different types of BPMN connecting lines. There is no obvious
reason to modify the semantics of the three types of connectors: Sequence
Flow, Message Flow, and Association.
o        The
clouds are valid through the Artifact extension mechanism, but can’t be
part of Sequence Flow or Message Flow.
o        The
Endpoint objects need to be clearly defined and the shape possibly changed.

Stephen A White/Irvine/IBM@IBMUS

09/10/2004 03:08 PM

To	David RR Webber <david@drrw.info>
cc	"'ebXML BP'" <ebxml-bp@lists.oasis-open.org>, Jean-Jacques Dubray <jeanjadu@Attachmate.com>, "'Monica J. Martin'" <Monica.Martin@Sun.COM>, Tony Fletcher <tony_fletcher@btopenworld.com>
Subject	Re: [ebxml-bp] [ebBP] 7/21/2004: Guiding Principles - BPMN and BPSS

Just to confirm what Tony was saying... 

BPMN was developed by a different working group than the one that developed
BPML. The make up of the Notation Working Group was mainly process modeling
tool vendors and some process modeling consultants. We were chartered with
two main goals: 1) develop a process modeling notation appropriate for
business people, and 2) develop a bridge to a process execution language.
Originally, that was BPML, but this was changed to BPEL when that seemed
more appropriate. We also were charted to be able to notate B2B situations.
We have some of this already and, thus, we are not likely far off in handling
what is in ebXML BP. 

We will have some comments and questions for your proposal soon.

-Steve

David RR Webber <david@drrw.info>

09/08/2004 06:04 AM

To	Tony Fletcher <tony_fletcher@btopenworld.com>
cc	Jean-Jacques Dubray <jeanjadu@Attachmate.com>, "'Monica J. Martin'" <Monica.Martin@Sun.COM>, "'ebXML BP'" <ebxml-bp@lists.oasis-open.org>
Subject	Re: [ebxml-bp] [ebBP] 7/21/2004: Guiding Principles - BPMN and BPSS

Tony,

Wow! 

I based my comment on a review of the spec' a while back when the BPMN
was in it - I had
no idea it was two teams!  I'll have to get the scoop from Howard
Smith 
on this sometime!!

Anyway - what is interesting is that I started with this same premise 
back last year - using
VisualScript to create an 'uber model' that could then be generated into
BPSS, BPEL, et al.

I worked with a couple of vendors too - and their own internal BPM 
engines script syntaxing,
and at first this seemed very promising.

Anyway - the wheel basically fell off this wagon when I realized that 
BPEL was diverging into
a 4GL for EAI vendors - with a arcane and complex behavioural execution
model - and more
importantly was not deterministic nor supported context or business 
transaction handling.
It's just attempting to solve a completely different problem set - one
is a train - the other a
hydrofoil triple-hulled boat design!

So - I've ended up focusing on BPSS as the core function set.   My
current thoughts envision
being able to use a modified set of BPMN symbols to represent BPSS 
concepts (where they
are close) and then being able to write out both BPSS XML instances and
then either
predicates and assertions for a Prolog engine like SWI Prolog / and / or
scripting for something 
like BEA, MQSeries or similar integration server.

The more things change they more they come full circle it seems!! ; -)

Cheers, DW.

Tony Fletcher wrote:

>Dear David, Jean-Jacques and others,
>
>Now I have caught up on this topic, I will try to keep up, for awhile
>anyway!
>
>Thank you, Jean-Jacques, for your message.  I will try to find
your mail
>with the proposal and take a look.
>
>Just a few corrections for David (as I participated in BPMI and BPMN
for
>about year ceasing at the back end of last year).
>
>BPMI (Business Process Management Initiative) is the organisation.
 BPML is
>the XML based business process description language they produced (a
year or
>more ago now).  No, it is not true to say that BPMN (Business
Process
>Management Notation) is the visual counterpart of BPML.  I was
not in at the
>beginning, but I understand that it was developed independently, with
a
>rather different set of people (some overlap I am sure) more based
on
>previous experience with process visualisation tools than any XML language.
>It used to have a section indicating how to approach representing a
BPMN
>diagram in BPML, however for the final version 1.0 of the BPMN specification
>that was removed and guidance on how to map to a BPEL4WS representation
>(Section 6) put in instead (yes, currently it is BPEL4WS, the proprietary
>version that was submitted to OASIS at the start of the WSBPEL TC -
when
>that TC produces a version of BPEL then the BPMN folks may produce
a new
>version of BPMN that shows how to map to it instead).  Overall
the design
>was aimed at visualisation tools rather than XML representation but
with the
>idea / hope that it would be able to map to several different 'popular'
XML
>BP languages such as BPML, BPEL, BPSS and the WfMC's XPDL.
>
>Best Regards     Tony
>A M Fletcher
> 
>Cohesions  (TM)
> 
>Business transaction management software for application coordination
>www.choreology.com
> 
>Choreology Ltd., 68 Lombard Street, London EC3V 9LJ     UK
>Tel: +44 (0) 1473 729537   Fax: +44 (0) 870 7390077  Mobile:
+44 (0) 7801
>948219
>tony.fletcher@choreology.com     (Home: amfletcher@iee.org)
>
>
>-----Original Message-----
>From: David RR Webber [mailto:david@drrw.info] 
>Sent: 07 September 2004 13:36
>To: Tony Fletcher
>Cc: 'Monica J. Martin'; 'ebXML BP'
>Subject: Re: [ebxml-bp] [ebBP] 7/21/2004: Guiding Principles - BPMN
and BPSS
>
>
>Tony,
>
>I thought that was BPMI?!?  The BPMN is supposed to be the visualization
>for that?
>
>Anyway - we can certainly use it to generate BPSS - and that is what
I'm 
>looking at - how
>to harmonize the BPSS diagrams I've been creating so far - that are
>loosely based on UML
>activity diagrams with BPMI - so that people used to creating BPMN
can 
>seamlessly
>cross over to using BPSS....
>
>Cheers. DW
>
>
>  
>

ebXML Example2.jpg