Realizability of Interaction Models Gero Decker ZEUS Workshop, Stuttgart 3 Mar 2009
Agenda <ul><ul><li>Why use the  interaction modeling style  (rather than the interconnection modeling style)? </li></ul></...
Process Choreographies <ul><ul><li>Inter-organizational integration requires interaction contracts </li></ul></ul><ul><ul>...
Choreography Modeling
Two Modeling Styles <ul><ul><li>Local ordering constraints </li></ul></ul><ul><ul><li>Distinction between control flow and...
Interconnection Models incompatible loops incompatible branching structures incompatible  sequence flow abstraction level ...
Interaction Models atomic interactions decision ownership global control flow dependencies complex interactions global loops
Interaction Petri Nets
Interaction Petri Nets Customer Manufacturer Reseller Payment Org
Realizability
Realizability – Two Dimensions <ul><ul><li>Communication model </li></ul></ul><ul><ul><ul><li>Synchronous (blocking send v...
Realizability – Existing Work <ul><ul><li>Fu, Bultan & Su:  Realizability  (2004) </li></ul></ul><ul><ul><ul><li>Projectio...
Fu, Bultan & Su:  Realizability  (2004) Traces = { [(A, τ ,A),(A,x,B),(B,y,C)], [(A, τ ,A),(A,x,B)]} Traces = { [(A, τ ,A)...
Now: Focus on Branching Structures
Role Projection:  Preserving Branching Structures
Role Projection:  Preserving Branching Structures
Role Projection:  Preserving Branching Structures
Open Issue: Proofs <ul><ul><li>Define  Full Realizability : There exist role behavior models the composition of which are ...
Open Issue: Subset of Behavior <ul><ul><li>If a choreography is not fully realizable then  </li></ul></ul><ul><ul><ul><li>...
Conclusion <ul><ul><li>Interaction modeling imposes challenge of realizability </li></ul></ul><ul><ul><li>Different flavor...
Questions? <ul><ul><li>Mail  [email_address] </li></ul></ul><ul><ul><li>Tool  http://oryx-editor.org   </li></ul></ul><ul>...
Upcoming SlideShare
Loading in...5
×

Realizability of Interaction Models

711

Published on

Published in: Technology
0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
711
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
8
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide

Realizability of Interaction Models

  1. 1. Realizability of Interaction Models Gero Decker ZEUS Workshop, Stuttgart 3 Mar 2009
  2. 2. Agenda <ul><ul><li>Why use the interaction modeling style (rather than the interconnection modeling style)? </li></ul></ul><ul><ul><li>When are choreographies realizable , i.e. there exist participants that collectively realize the specified behavior? </li></ul></ul><ul><ul><li>What impact do branching structures have on realizability checking? </li></ul></ul>
  3. 3. Process Choreographies <ul><ul><li>Inter-organizational integration requires interaction contracts </li></ul></ul><ul><ul><ul><li>Defined document structures / message types </li></ul></ul></ul><ul><ul><ul><li>Defined interaction behavior: constraints and obligations </li></ul></ul></ul><ul><li>“ A choreography defines the sequence and conditions under which multiple cooperating independent agents exchange messages in order to perform a task to achieve a goal state.” [W3C Glossary] </li></ul>
  4. 4. Choreography Modeling
  5. 5. Two Modeling Styles <ul><ul><li>Local ordering constraints </li></ul></ul><ul><ul><li>Distinction between control flow and message flow </li></ul></ul><ul><ul><li>MSC, BPMN </li></ul></ul><ul><ul><li>BPEL4Chor </li></ul></ul><ul><ul><li>Global ordering constraints </li></ul></ul><ul><ul><li>Interactions as atomic building blocks </li></ul></ul><ul><ul><li>UML Comm. Diagrams </li></ul></ul><ul><ul><li>BPSS, WS-CDL </li></ul></ul><ul><ul><li>Let’s Dance, iBPMN </li></ul></ul>Interconnection Models Interaction Models
  6. 6. Interconnection Models incompatible loops incompatible branching structures incompatible sequence flow abstraction level = messages
  7. 7. Interaction Models atomic interactions decision ownership global control flow dependencies complex interactions global loops
  8. 8. Interaction Petri Nets
  9. 9. Interaction Petri Nets Customer Manufacturer Reseller Payment Org
  10. 10. Realizability
  11. 11. Realizability – Two Dimensions <ul><ul><li>Communication model </li></ul></ul><ul><ul><ul><li>Synchronous (blocking send vs. failing) </li></ul></ul></ul><ul><ul><ul><li>Asynchronous (FIFO queue vs. standard buffer) </li></ul></ul></ul><ul><ul><li>“ Similarity” of behavior </li></ul></ul><ul><ul><ul><li>Complete behavior </li></ul></ul></ul><ul><ul><ul><li>Subset of behavior (proper termination vs. reachability constraints) </li></ul></ul></ul><ul><ul><ul><li>Equivalence notion (trace vs. bi-simulation vs. …) </li></ul></ul></ul>
  12. 12. Realizability – Existing Work <ul><ul><li>Fu, Bultan & Su: Realizability (2004) </li></ul></ul><ul><ul><ul><li>Projection: remove transitions + minimal deterministic automaton </li></ul></ul></ul><ul><ul><ul><li>Asynchronous, complete behavior (trace equivalence) </li></ul></ul></ul><ul><ul><li>Decker & Weske: Local Enforceability (2007) </li></ul></ul><ul><ul><ul><li>Projection: remove transitions + keep branching structure </li></ul></ul></ul><ul><ul><ul><li>Synchronous, subset of behavior (pattern-based + “blocked” transitions) </li></ul></ul></ul><ul><ul><li>Decker, Barros, Kraft & Lohmann: Desynchronizability (2008) </li></ul></ul><ul><ul><ul><li>Requires realizability for the synchronous case </li></ul></ul></ul><ul><ul><ul><li>Weak termination for the asynchronous case </li></ul></ul></ul>
  13. 13. Fu, Bultan & Su: Realizability (2004) Traces = { [(A, τ ,A),(A,x,B),(B,y,C)], [(A, τ ,A),(A,x,B)]} Traces = { [(A, τ ,A),(A,x,B),(B,y,C)], [(A, τ ,A),(A,x,B)]}
  14. 14. Now: Focus on Branching Structures
  15. 15. Role Projection: Preserving Branching Structures
  16. 16. Role Projection: Preserving Branching Structures
  17. 17. Role Projection: Preserving Branching Structures
  18. 18. Open Issue: Proofs <ul><ul><li>Define Full Realizability : There exist role behavior models the composition of which are bi-similar to the choreography </li></ul></ul><ul><ul><li>Theorem: A choreography is fully realizable if and only if the composition of the role projections is bi-similar to the choreography </li></ul></ul><ul><ul><li>Proof (idea): Show that the role behavior models must look exactly like they look if their composition is to be bi-similar </li></ul></ul>
  19. 19. Open Issue: Subset of Behavior <ul><ul><li>If a choreography is not fully realizable then </li></ul></ul><ul><ul><ul><li>Remove transitions from the composition that are not allowed by to the choreography </li></ul></ul></ul><ul><ul><ul><li> Imply which transitions in the role behavior models must be removed to achieve this removal in the composition </li></ul></ul></ul><ul><ul><ul><li>Try to find (reduced) role behavior models the composition of which shows a valid subset of the behavior of the choreography </li></ul></ul></ul>
  20. 20. Conclusion <ul><ul><li>Interaction modeling imposes challenge of realizability </li></ul></ul><ul><ul><li>Different flavors of realizability </li></ul></ul><ul><ul><ul><li>Communication model / similarity of behavior </li></ul></ul></ul><ul><ul><li>Role projection as “best guess” for local behavior </li></ul></ul><ul><ul><li>Branching structures important if local choices to be considered </li></ul></ul><ul><ul><li>Role projection must reflect branching structures </li></ul></ul><ul><ul><li>Unifying theory still missing  Niels? </li></ul></ul>
  21. 21. Questions? <ul><ul><li>Mail [email_address] </li></ul></ul><ul><ul><li>Tool http://oryx-editor.org </li></ul></ul><ul><ul><li>Blog http://bpmn.info </li></ul></ul>
  1. A particular slide catching your eye?

    Clipping is a handy way to collect important slides you want to go back to later.

×