I once asked a question in Use of Aris models for Guided Procedures or VC possible. I never really thought about the reasons behind his answer. I was curious what the long answer might have been but I forgot about it and started exploring other topics.
Recently, I was reading the BPEL4People specification (supported by IBM and SAP) which is an optional extension to BPEL 2.0 that would standardize human tasks in a BPEL process, and I started to think about my old forum post.
I asked myself what might the long answer for my question look like.
A Caveat: I am not a BPEL expert who know the specification (this Blog reflects BPEL 1.1 more than BPEL 2.0) inside and out. So, if this blog contains any errors concerning BPEL, then please yell and scream in the comments. As mentioned in other blogs, my desire is to start others thinking and discussing these issues.
Caveat 2: I am not a SAP Insider, so I have no idea what the internal product managers for GP are thinking/doing about this issue. I do, however, have a Java decompiler and have found some intriguing Java classes in the trial version of the 2004s WAS. So, I know that somebody at SAP is looking at this stuff.
Lets first look at my motivation for asking the question. I was examining the gap between the work of the BPX at a design level and the technical implementation of this design. SAP Enterprise Modeling Applications by IDS Scheer, so I decided to use it as my starting point. I knew that ARIS could export BPEL and that there was an interface for Exchange Infrastructure (XI)., so I started looking for information regarding 1) ARIS- Guided Procedure (GP) interaction and 2) BPEL GP interaction. I didnt find anything, so I posted my forum question.
After reading the BPEL4People specification, it became clear why a complete mapping between BPEL and GP is unlikely.
GP ≠ XI
As I mentioned above, XI has the ability to import BPEL. This means that in general SAP has enough technical knowledge regarding BPEL to provide a BPEL import for GP if this was technically possible. The reason that this import functionality doesnt exist in GP is that BPEL and GP dont really match well. Furthermore, XI is not the same product as GP.
If you tried to create a table with the appropriate mappings for BPEL and GP (just a quick attempt), you might end up with a table like this:
BPEL | GP |
process | process |
sequence | sequential block |
flow | parallel block |
assign & copy | set parameters for actions |
variables | parameter in the process context |
Obviously, there are a variety of missing GP elements / mappings. For example:
Thus, from a technical perspective, a possible BPEL GP mapping would leave much to be desired.
BPEL is a standard
BPEL and BPEL4People both intend to be standards that are product-independent this means that the technology of a particular product should not be included/promoted at the expense of other products / technologies. GP is part of the SAP NetWeaver and includes special features that take advantage of other SAP products/ technologies (BSP, CAF Core, Web Dynpro, etc.). This is also understandable from the perspective of SAP. If you look at GP, you notice that many of these special features are based on the Callable Object level the diversity and power of these objects is, in my opinion, what really makes GP stand out. Thus, although a partial integration of BPEL in GP may be possible at the Process, Block and Action level, a standard will probably be unable to include those SAP-specific objects at the Callable Object level.
Effects of the Developmental Process
In general, the relationship between GP and the design environment of the BPX is still largely undefined. It is obvious that processes that involve human interaction should in all likelihood be implemented using GP. Does this mean, however, that the BPX can change processes in GP that he designed using an external tool (for example, ARIS)? If this is permitted in GP, there must be round-trip possible from GP to this external tool; otherwise, there will be a discrepancy between the realization in GP and the design in the external tool. Or do these environments represent different design platforms: a theoretical and a technical design? Irregardless, this question is still largely unclear and the answers might even be different based on organizational context.
This discussion has an impact of the use of BPEL in GP, because it impacts both the depth and basic characteristics of the relationship (who has the lead design tool or GP?) (Or are the tool environments placed on equal footing, meaning that changes in both environments must be reflected in the other?) that must reflected in a potential BPEL<->GP implementation.
Based on the realization that a complete BPEL-GP interaction is unlikely what are the repercussions of this supposition?
Thus, the long answer is still basically NO but at least I now know why.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
5 | |
5 | |
5 | |
4 | |
4 | |
4 | |
4 | |
4 | |
3 | |
3 |