Comments on: Process Model Portability - Does Anybody Care? http://www.brsilver.com/wordpress/2007/11/28/process-model-portability-does-anybody-care/ Bruce Silver's blog on business process management Fri, 05 Sep 2008 13:59:36 +0000 http://wordpress.org/?v=2.5.1 By: bruce http://www.brsilver.com/wordpress/2007/11/28/process-model-portability-does-anybody-care/#comment-2953 bruce Wed, 28 Nov 2007 23:12:49 +0000 http://www.brsilver.com/wordpress/2007/11/28/process-model-portability-does-anybody-care/#comment-2953 Warwick, Supporting BPMN is more than using rectangles and diamonds. It means expressing all of the flow semantics using the notation as described in the BPMN spec. This has nothing to do with vertical swimlanes or curved sequence flows or other things that give ALBPM its distinctive "look". I think BEA actually agrees with this and will gradually move its notation to true BPMN. It would actually give them an advantage, because their engine is one of very few that can handle everything in the BPMN spec. On the portability front, I am working with the XPDL 2.1 guys to try to define a "model portability conformance class". The idea is that a conformant abstract model - abstract meaning essentially what prints in the BPMN diagram, not the data or other implementation attributs - would be portable between any 2 compliant tools. That would be HUGE! Portability conformance would apply to a particular instance document Process.xpdl serialized in XPDL 2.1. My idea is this: 1. Define ClassNFilter.xslt, which applied to Process.xpdl generates ClassNProcess.xpdl. The resulting instance document contains only the elements and attributes required to be portable. This is easy. I've already done it. 2. Define ClassNValidator.xslt, which contains the validation rules. Applied to ClassNProcess.xpdl, it generates a list of errors based on the required elements and attributes and links between them. This is a bit of work, but doable. If there are no errors, we say Process.xpdl is ClassN model portability conformant. There may be multiple conformance classes, depending on how much of BPMN must be supported, e.g. which event types, or compensation, etc. A modeling tool asserting that it is ClassN compliant means that it can import any ClassN conformant model instance created in another tool and understand its meaning... at least as far as the elements and attributes contained in ClassNProcess.xpdl. The other elements may not be imported successfully but that would not forfeit compliance to ClassN portability. My hope is that a significant number of modeling tools would be base class compliant, and those that weren't would feel pressure to do so. I am confident this scheme would work, although it is a lot of work to create ClassNValidator.xslt (mostly a list of XPath expresions for the rules). I'm not sure I've convinced Robt Shapiro and the other XPDL guys of the value yet. Warwick,
Supporting BPMN is more than using rectangles and diamonds. It means expressing all of the flow semantics using the notation as described in the BPMN spec. This has nothing to do with vertical swimlanes or curved sequence flows or other things that give ALBPM its distinctive “look”. I think BEA actually agrees with this and will gradually move its notation to true BPMN. It would actually give them an advantage, because their engine is one of very few that can handle everything in the BPMN spec.

On the portability front, I am working with the XPDL 2.1 guys to try to define a “model portability conformance class”. The idea is that a conformant abstract model - abstract meaning essentially what prints in the BPMN diagram, not the data or other implementation attributs - would be portable between any 2 compliant tools. That would be HUGE!

Portability conformance would apply to a particular instance document Process.xpdl serialized in XPDL 2.1. My idea is this:
1. Define ClassNFilter.xslt, which applied to Process.xpdl generates ClassNProcess.xpdl. The resulting instance document contains only the elements and attributes required to be portable. This is easy. I’ve already done it.
2. Define ClassNValidator.xslt, which contains the validation rules. Applied to ClassNProcess.xpdl, it generates a list of errors based on the required elements and attributes and links between them. This is a bit of work, but doable. If there are no errors, we say Process.xpdl is ClassN model portability conformant.

There may be multiple conformance classes, depending on how much of BPMN must be supported, e.g. which event types, or compensation, etc.

A modeling tool asserting that it is ClassN compliant means that it can import any ClassN conformant model instance created in another tool and understand its meaning… at least as far as the elements and attributes contained in ClassNProcess.xpdl. The other elements may not be imported successfully but that would not forfeit compliance to ClassN portability. My hope is that a significant number of modeling tools would be base class compliant, and those that weren’t would feel pressure to do so.

I am confident this scheme would work, although it is a lot of work to create ClassNValidator.xslt (mostly a list of XPath expresions for the rules). I’m not sure I’ve convinced Robt Shapiro and the other XPDL guys of the value yet.

]]> By: kewbss http://www.brsilver.com/wordpress/2007/11/28/process-model-portability-does-anybody-care/#comment-2948 kewbss Wed, 28 Nov 2007 21:56:26 +0000 http://www.brsilver.com/wordpress/2007/11/28/process-model-portability-does-anybody-care/#comment-2948 Bruce, A couple of comments ... 1) BEA Aqualogic BPM Suite already provides a BPMN modeling mode although I'm not sure if it is totally compliant with the standard. On thing that is different about it is the use of curved rather than straight flow lines, but that's probably not contrary to the standard is it? 2) I'm working on a project at the moment where model transportability is important. This is the first BPM project for the Client and while they are making all the right moves to select a suitable BPMS, they understand that each has its own strengths and weaknesses. To mitigate the technology selection risk, the first phase of the project involves only modeling, simulation and pilot implementation. If the initial BPMS were to fall short, they would re-evaluate the technology requirements and look for a more suitable platform. Obviously they would not want to re-develop the model in the new tool. Warwick Bruce,

A couple of comments …

1) BEA Aqualogic BPM Suite already provides a BPMN modeling mode although I’m not sure if it is totally compliant with the standard. On thing that is different about it is the use of curved rather than straight flow lines, but that’s probably not contrary to the standard is it?

2) I’m working on a project at the moment where model transportability is important. This is the first BPM project for the Client and while they are making all the right moves to select a suitable BPMS, they understand that each has its own strengths and weaknesses. To mitigate the technology selection risk, the first phase of the project involves only modeling, simulation and pilot implementation. If the initial BPMS were to fall short, they would re-evaluate the technology requirements and look for a more suitable platform. Obviously they would not want to re-develop the model in the new tool.

Warwick

]]>