BPMS Watch

On a Reference Model for BPM 2.0

August 10th, 2006

Ismael posts an interesting reader request on IT|Redux addressed, it seems, to both of us:

Ismael, Bruce:

Do you think maybe it’s time for experts such as yourselves to get together and establish a standard model for BPM 2.0 in simple, concise and hopefuly universally applicable terms that can be ported throughout various functions and industries?

My work is in the supply chain arena, where about 10-15 years ago there was a similar debate regarding what constitutes a supply chain, which business function was responsible for what portion of the chain, what were some characteristics unique to the supply chain, but standard enough that could be used across all industries, etc… you get my point.

On that background, a number of experts and major clients in the supply chain arena got together and formed the Supply Chain Council, with the mission of developing a standard model, performance metrics and best in class practice, which they called the Supply Chain Refernce Model, or SCOR.

This maybe be to a certain extent what BPM 2.0 needs — a standard model, the basic tenets of which all experts agree on. A model that is relevant to all parties (Business and IT), yet simple enough that it can be easily understood by all involved, so that its use is being promoted by the debate, and BPM’s adoption rapidly expanded.

SCOR is such a model in my area of expertise, and I can tell you that by its development and use modern supply chain concepts were easily and widely adopted by practitioners.

Just a thought.

-Ryan Armasu

He goes on to relate his experience with BPMI and his personal reaction to Ryan’s idea, which is worth reading.  I’m wrestling myself with what such a reference model should include, and how it relates to other standards out there, like BPMN and BPEL, or even Gartner’s BPMS functionality checklist to qualify for the magic quadrant.  Individually and in combination, these three de facto standards have not provided clarity around what BPM is, much less the much narrower notion of BPM 2.0. 

For example, take BPMN.  Is it a design language or just a drawing notation?  OMG won’t say.  I’m not sure if they even know.  Eventually it may have a metamodel, but it still has no methodology.  It has nowhere to put simulation parameters used by virtually every BPMN-based modeling tool.  It does not distinguish between core elements that must be supported by every compliant tool and other elements that are optional.

For example, take BPEL 2.0.  Is portability a goal?  What about human tasks?  Subprocesses?  These are not included, just vague white papers that surfaced a full year ago with not a peep since.  BPMN devotes half the spec to BPEL mapping, but as Assaf Arkin and others have pointed out, hasn’t aligned its concepts with BPEL sufficiently to allow such a mapping unambiguously.

For example, take Gartner’s checklist.  It’s fine as a list of functions that an executable process design should support, but when the suite is defined as a set of SKUs from a single vendor it starts to make less sense.  Content management?  Business rules?  In SOA, aren’t these just services?

Ismael’s starting point is the combination of BPMN and BPEL.

If I were to create a standard model for BPM 2.0, I would start by defining what it is, without trying to re-invent the wheel, and the quickest path toward this would be to start from the BPMN and BPEL specifications. Take BPMN and BPEL, figure out what this gives you, then describe a set of benefits that business and IT users would get out of it.

I don’t agree, because the only product that combines them today is Intalio.  Maybe when Oracle integrates ARIS and adds their special sauce, they might qualify.  But who else?  So BPMN+BPEL lacks critical mass.  I prefer something like that but more “abstract” or at least more technology standards-neutral.  My abstraction of BPMN would be a business-oriented modeling notation that supports both analysis and basic executable design, and includes semantics of events and choreography in addition to orchestration.  My abstraction of BPEL would be an execution language that supports the general notion of “service orchestration” including fault handling and business transaction recovery, and I’d have to throw in subprocesses and human tasks as well. 

Now at least we’ve got the chance of critical mass: IBM, Lombardi, BEA, Oracle, Cordys, Intalio…  Within this technical framework, BPM 2.0 needs to add a general methodology and best practices.  When you’re using BPM 2.0 to model and build a process, what’s the order of activities?  What functions are provided by business analysts, process analysts, and developers, using which tools, and what skills are implied by those titles anyway?  How does the model drive the implementation?  What elements should be added by IT, and to what degree should these be surfaced in the business view?   What does this top-down, bottom-up, middle-out business really mean?

Things like portability of the design are orthogonal to BPM 2.0.  Maybe BPMN 2.0 will be the portability standard, maybe BPEL 3.0, or maybe even XPDL 3.0.  BPM 2.0 shouldn’t be about portability, but about how business and IT collaborate to model, design, and manage business prcesses.

Entry Filed under: BPM