Maybe I’m just a hopeless daydreamer, but I think that if all standard makers were really transparent about what I’ll call “business intentions”, we could conceive the idea of several co-existing standards. But as I’m aware of the unlikeliness of this happening, I’m just going to toss a few guesses on both sides in this case.

Maybe the ultimate audience of a standard is the whole industry, but it needs a vehicle audience to get there. In this case, for OMG, that vehicle are those major vendors who develop costly BPM tools and suites, such as Lombardi, Axway, Adaptive, MEGA International among others, who also happen to be authors of the BPDM specification. It’s like creating the moon to be the first ones to walk over it, but that’s the way it goes and has always been. For example, I’m not sure about the details on how UML reached its popularity, but the pricey CASE tools had their part, being its vendors the ones that wrote the spec. However, will the rules of the game remain the same?

Much has been said of the rapid growth of the Eclipse Foundation, but little has been mentioned about its importance to OMG standards. However, something has been said about Eclipse becoming the builder of Reference Implementations for OMG standards, and even the role of implementations shaping the standards making process [1]. And we ought to remember that Eclipse is the most successful OSS experiment to this date, along with Apache. Now we see Intalio has built a BPMN modeler, and it’s not just another plug-in; it’s part of the SOA Tools Platform, an Eclipse project. And if things work out as they are described in [1], Intalio’s schema would become the industry’s reference implementation for BPMN.

It is sad to make up teams that gather disjoint efforts only to have two and only two parties in a senseless battle. What sense does it make to create coalitions when everybody is after the same goal, but only their ways are different? In any case, we’ll be all watching what will work and what not, which is to be told really soon.

[1] Eclipse Modeling Project and OMG Standards. Version 1.0, September 25, 2006.

PSL (Process Specification Exchange) is an emerging ISO standard that complements ISO AP233 (Product Data Exchange) and EDI. BPDM is based on PSL — axiomatic supplemented by an ontology — which will enhance BPMN and eventually replace BPEL. The current combination of BPMN and BPEL is good for the choreography of processes within an enterprise. BPDM will support the orchestration of processes for backward and forward integration into suppliers and customers. This round out the management of supply chain end to end.

I am sure that SysML will be used by the Defence Department of Australia to implement its Enterprise Architecture, which includes business modelling. I just connected up with an EA practitioner of DoDAF. He led the effort to deliver an architecture that ensures the inter-operation of technologies deployed by the US, UK, Canada, Australia, and New Zealand when their arm forces participate in a joint-mission.

I believe the above gives you an adequate explanation in a nutshell.

Regards
 -Francis

I’m practically nobody, just a person with more than 25 years of experience in management consulting, and an IT-literate end-user with over 40 years of experience in dealing with people in IT. I don’t participate in standard making activities, I just follow the standards adopted by nationally or internationally accredited standard bodies, ISO in particular, and who can really drive standards.

My take is that BPM will be another fad in a few years, just like Structured Methodology in the 70s, Information Engineering in the 80s, and Object-Oriented Approach in the 90s. These movements solve a few problems, but never solve the age-old problem of the communication gap between business and IT communities! One thing I am sure of is that the business community always wins, as it holds the purse’s strings! IT trying to drive the business has never worked and never will!

I don’t know which OMG standard will win, BPMN+BPEL or BPDM! I believe that the one that hooks up easily to EDI (ISO & ANSI), AP233 (ISO & NIST), and PSL (ISO & NIST) will win! A working standard doesn’ t always win. If it does, there is no need for new or improved standards! There is no standard or technology that can solve every problem under the sun. When there was an Activity Diagram in UML, why did we need a BPMN standard? UML is already an ISO standard, while BPMN is not. Nevertheless, UML has been an IT driven standard. Thus, SysML, a profile of UML, was born and incorporated into several ISO standards.

As far as I am concerned, the debate on which standard is better, BPMN+BPEL vs BPDM, is only an academic exercise. Let us see which one UPDM, the most influential UML profile task force, will adopt.

Best regards
Francis Ip

That makes sense, as long as the work done remains 100% focused on BPMN, and does not include anything specific to XPDL. I think this would be a great step forward. How about you write about this idea on your blog and invite Keith and I to take the lead on the effort?

Best regards
 -Ismael

I’ll be happy to continue to talk to you about it. At first glance it didn’t seem fundamentally different from XPDL (if you strip out XPDL’s graphics-related info and change the element names). You should take a look. If you wind up agreeing, since the standards game is at least 50% political, you might think about teaming up with Keith Swenson so there is just one alternative serialization (besides BPDM), not two.

-Bruce

I could not agree more. BPDM at present time is overkill. Actual support for PSL would be interesting, but it’s not available yet, and it would require changes to the BPMN specification in order for it to be really useful. And a simpler serialization format for BPMN is what people really need, not a new meta-model from the sky that virtually nobody will fully implement.

If BPDM gets any support at all, it will be for a small subset of what it provides, and the meta-model is so complex that using it for interchange between products developed by different vendors will be virtually impossible. Recent discussions on the Eclipse mailing list from people who know a couple of things about meta-models go in this direction. If even the most technical people do not care about such a meta-model, why should less technical people such as business analysts and process analysts pay any attention at all?

At this point, I would strongly encourage you to provide as much feedback as possible on the simple meta-model and schema we developed for BPMN and donated to the Eclipse Foundation. I believe there is an opportunity for these to be adopted by pragmatic vendors that have a vested interest in the success of BPMN, the only process modeling standard that really matters.

Best regards
 -Ismael

Who is this “Francis Ip” anyhow, and how have my comments on BPDM offended him so? Or exactly, what is it I need to “retract” about process semantics? He has never commented on or tracked back to my blog, just hides behind IT|Redux…

Anyway, for the record, while you are free to characterize my comments in Intelligent Enterprise any way you like, what I actually said about BPDM was this:

“I’ve actually seen a draft of BPDM and see no signs of a BPMN schema. Actually I found the thing near-incomprehensible; there was something about MOF and XMI but not a schema. It made me wonder whether BPDM would actually include a schema for BPMN, or just some kind of production rules that ensure conformance to the BPDM metamodel… If OMG does not publish a BPMN schema, I see more consternation in BPM-Land and a second chance for XPDL to get it right. If BPDM produces a schema and a list of must-support BPMN semantics, then I predict next year [that it will kill XPDL].”

So the near-useless part (to me, anyway) is BPDM without a schema. A couple days ago they did publish a schema, of sorts. Well, it’s a set of xsd files, although XMLSpy couldn’t fully open them (apparently they don’t conform to the complexType schema in the xsd namespace), and there are no root elements or other things that usable XML schemas contain. And OMG didn’t even publish an instance document along with a diagram so ordinary people might make sense of it. Maybe they’ll get there eventually; they did say the schema needed some “fine tuning” still.

Notwithstanding the scurrilous blathering of Mr. Ip, I feel safe in saying the big demand from the BPM community regarding BPMN has always been for a standard interchange format, i.e. an XML schema and the most basic portability standards, not for a more formal metamodel. Steve White has always maintained that BPMN 1.0 already provides a metamodel, but it ‘s just “informal” in the narrative, and I agree. OK, it’s vague on a few things, but the bigger problem is BPMN does not require “compliant” tools to support its most basic constructs, like intermediate events (absolutely important for portability).

So at this point, I’m not yet pro or con on BPDM. I’m just waiting for something useful to emerge.

-Bruce

Isn’t BPDM a technology as well? Show your specification to anyone — especially business folks — and ask them if that is not technology. I would love to see their reaction. Just because you call it a meta-model does not give it any special status, or make it any closer to the absolute truth that you’re looking for. There can be as much semantics in a well-written XML schema as in a UML Diagram. It’s all in the eye of the beholder.

I’m sorry, but for all it’s fancy claims, MDA is just yet another technology, and it happens to be one so overly complex that it plainly does not work. Never did, never will. I might be naive, but I like simpler solutions, and the BPMN+BPEL duo is one that works today. When you get something that works as well, for as many business scenarios, please show it to me and I might change my mind.

Now, I’m glad to read you clearly state that BPEL does not matter. This is where we differ, and this is the problem I have with BPDM’s charter. The industry at large has agreed on a common process execution model, and the BPDM working group is ignoring it. That’s fine by me, it just makes your effort all the less interesting.

Best regards
 -Ismael

You are very correct that BPMN and BPEL are “here today”, but what about the last 2000 technologies and notations there were “here today”? Why was BPEL done when we have so many ways to execute a process already? You can, of course, do everything in Java or Cobol or WfMC standards that you can do in BPEL!

The parade of technologies keeps coming and I understand it is in the best interest of vendors, like you, to keep it coming and to get everyone to believe in the latest silver bullet. But, some users want out, they don’t want their business processes tied up to the latest technology. They just don’t believe it anymore.

In addition, enterprise solutions are not just BPM, or SOA, or the DBMS — solutions need to tie these paradigms together in a coherent fashion. As each of the stovepipes fosters yet another community, set of products, etc., it becomes harder, not easier, to assemble real solutions that need a bit of everything. The technology centric “stack” we have is 50 times harder to use than what it replaced — and more are coming next week.

So this is where MDA fits in, and fits in by describing models of the enterprise that can fit together, and do so without dependence on the “execution language” or yet another diagram. The model captures the full life-cycle from business needs to execution model — based on the semantics, not the technology.

Since we have the machinery to execute this on top of the last 50 or so technology revolutions, it will not take long for you to be able to go from model to solution with BPDM — leveraging BPEL, if that is your technology choice. But, when the fads change, so will our mappings, leaving you in the dust.

On the other hand, if you want to say, it’s all about the XML in the schema — I don’t see how the marketplace will benefit from you proposing yet another process schema, instead of getting on board with BPDM and using a standard — and offering constructive suggestions if there are issues. Because the simple fact is, despite the FUD, you can serialize and de-serialize BPMN semantics using BPDM — we just don’t need another one.

If the BPDM group actually manages to release a public specification for a meta-model that supports BPMN and PSL, I will certainly take a very close look at it. Until then, what is being discussed here is more akin to science fiction, and while I like the genre, I have better sources for it.

Best regards
 -Ismael

If that is the goal, then why not adopting BPEL as the underlying execution language? Why should we make everything more complex than it should be? You say you support PSL, but you did not include such support in the specification. What is the value then? All I can see right now is yet another pie-in-the-sky project that never delivers anything useful. As far as I am concerned, BPMN and BPEL are real, available today, and all we need is a way to go from one to the other, which is exactly what the BPMN schema we developed does.

Best regards
 -Ismael

It is nice to know that the BPDM standard is based on rigorous semantics, which consists of axioms and ontology as defined in PSL-ISO18629. Unlike some standards that use XML schemas as semantics. A schema is basically a description of the structure of a document, which has no contextual meaning. When a schema complies with the XMI standard, the meta data is well defined and the document can be exported and imported by different software tools. In the absence of an ontology, BPMN has only implicit semantics, which one can interpret in anyway s/he likes.

I don’t know whether you are aware of UPDM or not. Just in case, it is an OMG profile task force that defines modelling standards to support DoDAF (US) and MODAF (UK). SysML and UML 2.1 are currently the preferred standards supplemented by IDEF0 (used in STEP EXPRESS-ISO10303-11) and BPMN. There seems to be a movement in enhancing the BPMN, which will eventually replace the Activity Diagram in UML In general, it is wise to observe what US DoD and UK MOD are doing in terms of driving standards and adoption of technologies. Major vendors like IBM, ORACLE, Microsoft, Telelogic, SAP, and the like pay special attention to these two largest technology users in the world. Sun Microsystems is the exception! It should also be noted that military forces of NATO, Canada, Australia, and New Zealand follow more or less the same set of standards. The primary objective is to ensure that the systems (e.g. Command, Control, Computer, Communication, Intelligence, Surveillance, & Reconnaissance) used by all these forces will interoperate, in a plug-and-play fasion, when they participate in a joint mission.

I speculate that once the BPDM standard is released, UPDM will most likely incorporate it into the profile. Moreover, it won’t be long before DoD will mandate the compulsory EDI with RFID as its acquisition standard, which will track an item from its producer all the way to the end user and its eventual disposal.

It would be interesting to see what Dr. Bruce Silver will say about process semantics! There is a possibility that he needs to retract what he has said before. He seems like a partial replacement of Dr. James Martin, the IT guru and ex-IBM research fellow, in the 60s, 70s, and 80s.

Keep up the good work.

Best regards
 -Francis

You will be happy to note that PSL was and remains a major part of BPDM. Conrad Bock of NIST was an integral part of the team and was able to ground the BPDM process semantics in PSL — this is why we have such confidence in the semantic integrity. We didn’t make that part of the specification available since many people don’t understand it — but the work has been done and I think can be made available. As for the other bodies and activities, I certainly agree — we need more cohesion in our standards supporting the enterprise.

Regards,
Cory Casanave

I believe that the PSL (Process Specification Language) for process specification exchange should be one of BPDM activities. NIST is working with ISO on such a specification that supplements AP-233.

The prime interest of Dr. Bruce Silver — an independent analyst — is to run seminars on anything related to BPM to make money. Late delivery of the BPDM specification doesn’t give him the opportunity to make more money.

The combination of BPMN, BPEL, BPMS, and BPDM have a long way to go in making Enterprise Process Management a reality! In addition to OMG, we need to pay more attention to other accredited standard bodies (e.g. IEEE, ISO, etc.), as well as power users of standards (e.g. US DoD and UK MOD).

Best regards
-Francis Ip

The dominant forces in producing BPDM were BPMN vendors, business and process modeling practitioners, business process modeling vendors, and government representatives — not “legacy workflow vendors” at all. There is no plot and no fight, BPDM fully embraces and supports BPMN, and provides added value as we will discuss below. Technically BPDM is a superset of BPMN, with added support for SOA — far from the centrally managed style of the old workflow systems. Perhaps we can keep our debate at a higher level. The observation that BPDM is complex to understand has more merit — but let’s look at the context and why.

Process semantics: Some other approaches to process “exchange formats” are little more than pictures in XML. While this provides interchange of the diagram, it fails to exchange a process specification. One of the reasons that BPDM took so long is that nailing down what is really being said in process models is harder than it may appear, the team had the advantage of being able to work with the BPMN team directly, to iron out the real semantics — and there were a lot of arguments and disagreement about what they really meant. BPDM captures the result of this and provides a way to exchange the process semantics, not fuzzy pictures or partial thoughts. Semantics like this does take more time to grasp — but the result is real sharing of processes, not lock-in to different vendor’s interpretations and extensions.

Integrating processes and SOA: Process orchestration is a valuable tool in the enterprise toolbox, but not the only one. Process is part of the business model that includes services, organizations and resources. The SOA perspective in particular has been fully integrated, so that the process and the services offered and used are well integrated into a coherent enterprise perspective. BPMN can’t fight SOA, it needs to join it. Orchestration and SOA are just different ways to look at process. BPDM provides a bridge between those views. That bridge can be ignored by those who only want to understand orchestration, but we don’t think it will be ignored by the users who need both.

So this additional “complexity” of actually understanding what processes mean and how they relate to services and other processes has an impact, but that impact is primarily the responsibility of the builders of process tools, frameworks and infrastructures — perhaps having them take the time to understand the semantics behind their diagrams and runtimes is a good thing, I’m sure they are more than capable of grasping it.

There will be an open-source Eclipse based implementation of BPDM on ModelDriven.org as well as a schema for interchange — those who just want to know how to exchange BPMN models can use this, without the complexity. Perhaps we can also use this forum for an informed debate on the best way to represent process and provide the best approach to our communities and organizations.

-Cory Casanave
Model Driven Solutions

Thanks for the kind words, I’m sure my team will appreciate.

Best regards
 -Ismael

I have been following the development of the BPMN since its creation at BPMI.org before the merger with OMG, watching how different implementations have appeared on the scene. I’m particularly glad to see a BPMN modeler built on top of EMF, which implies the formal schema was developed with eCore, even when the current specification of BPMN doesn’t demand the notation to be MOF compliant. This is brilliant!

A few months ago in Washington DC, I met the CEO of one of the companies that authored the BPDM specification, who I was contacting to get some ideas for my senior undegraduate project. I knew I wanted something related to business process modeling, but still couldn’t really grasp something useful to work on. The suggestion I got was to browse a few generic process models on the web, and formalize them for MDA according to the BPDM specification. When I got my hands on it, I was overwhelmed at the complexity of this abstract proposal, and had to give up this idea for my project.

My project now consists in another effort to bring down the semantic gap between IT and business by following the MDA principles, and finding heuristics to specify the most relevant aspects of a SOA, always starting from a business process, of course.

Luckily, I remembered Eclipse had a brand new BPMN modeler, and decided to look into it. The schema that your team has built is as simple as it is great, and best of all, it works wonderfully. I have already contacted your team about joining as a developer, and started looking at the code. Hope I can start contributing soon!

Regards,
Juan Cadavid