Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
130905 francis palma - detection of process antipatterns - a bpel perspective
1. Detection of Process Antipatterns:
An BPEL Perspective
Francis Palma, Naouel Moha, and Yann-Gaël Guéhéneuc
MRI-BP 2013
Ptidej Team, École Polytechnique de Montréal, Canada
Latece, Université du Québec à Montréal, Canada
2. Context
Context
• Service Oriented Architecture (SOA)
Motivation
Related Work
• Service-based systems (SBSs)
• Business Process Modeling Notation (BPMN)
• Business Process Execution Language (BPEL)
Approach
Experiments
• Service orchestration using control logics
• Design patterns and Antipatterns
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
2 of 21
3. Context
Users
Client application
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
4. Context
Users
Client application
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
BPMN model
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
5. Context
Client application
Users
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
BPMN model
Approach
Experiments
Conclusion
Early design
anomalies
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
6. Context
Client application
Users
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
BPEL process
BPMN model
Approach
Experiments
Conclusion
Early design
anomalies
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
7. Context
Client application
Users
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
Transformation errors
BPEL process
BPMN model
Approach
Experiments
Conclusion
Early design
anomalies
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
8. Context
Client application
Users
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
Transformation errors
BPEL process
BPMN model
Approach
Experiments
Conclusion
Early design
anomalies
September 10, 2013
Conceptual
mismatch
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
9. Context
Client application
Users
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
Evolution
Evolution
Transformation errors
BPEL process
BPMN model
Approach
Experiments
Conclusion
Early design
anomalies
September 10, 2013
Conceptual
mismatch
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
10. Context
Client application
Users
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
Evolution
Evolution
Transformation errors
BPEL process
BPMN model
Approach
Antipatterns!
Experiments
Conclusion
Early design
anomalies
September 10, 2013
Conceptual
mismatch
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
11. Context
Users
Client application
BPEL Engine
Web services
Context
messages
BPEL Process
messages
Motivation
Related Work
Evolution
Evolution
Transformation errors
BPEL process
hard to maintain
and evolve
BPMN model
Approach
Antipatterns!
Experiments
Conclusion
September 10, 2013
bad quality of
service
poor design
Early design
anomalies
Conceptual
mismatch
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
3 of 21
12. Context: Process Antipatterns
Context
Start
Deadlock!!!
Motivation
Related Work
(a) Deadlocks Through Decision-Join Pairs
Approach
Experiments
Conclusion
September 10, 2013
(A,B)
D
(A,B,C)
(b) Dangling Inputs and Outputs
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
4 of 21
15. Motivation: Why Antipatterns
Context
Motivation
• Antipatterns:
- From wrong design decisions to poor solutions
• Poor solutions result:
Related Work
Approach
- bad quality of service (QoS)
- less maintainability, evolvability etc.
• Detect antipatterns within processes
Experiments
• Improve design and QoS
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
5 of 21
16. Motivation: Why BPEL ?
Context
Motivation
Related Work
• BPEL processes are off-the-shelf entities, BPMN is more
documentation related
- transition often broken, not symmetric
• Ideologically, antipatterns in BPMN and BPEL processes represents
same semantics, but some syntactic differences exist
• Antipatterns in models (BPMN) already got much attention
in the literature:
- target audience: business analysts
Approach
Experiments
Conclusion
September 10, 2013
• Conceptual mismatch between BPMN and BPEL
- differ in control flow support, e.g., loops, splits, joins, etc.
• Transformation errors and early design errors:
- translation, adaptation, and–or implementation errors
- errors by business analysts
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
6 of 21
18. Related Work (1/2)
Context
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Model Antipatterns:
- Onoda et al. (1999): catalog of 5 deadlock patterns
- Persson et al. (2006) and Stirna et al. (2009): 6 process patterns
and 13 process antipatterns
- Koehler and Vanhatalo (2007): 14 structural antipatterns in process models
- Trcka et al. (2009): formalized 9 process antipatterns using temporal logic
Detection of Model Antipatterns:
- Gruhn and Laue (2010): heuristic-based approach for discovering
problems in BPMs
- Laue and Awad (2010): visually represented process antipatterns
Process Patterns:
- Wohed et al. (2002): analysed BPEL4WS based on workflow and
communication patterns
- Aalst et al. (2003): discussed 26 control flow, branching-synchronization,
and structural patterns
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
7 of 21
19. Related Work (2/2)
Identified gaps from the literature:
Context
Motivation
Related Work
Approach
Experiments
- Antipatterns and detection approaches were considered
only for BPMN models
- No other formal specifications for process antipatterns
except the one by Trcka et al. (2009)
- Various quality aspects (e.g., availability or response time
of Web services) were not considered
- No automatic detection approach for BPEL process
antipatterns until now
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
8 of 21
21. Solution towards detection
Context
Motivation
Related Work
Approach
Experiments
• We propose to:
- specify process antipatterns using classical Rules of
Inference
- define a concrete approach
• We perform a small detection experiment
- seven antipatterns, three example BPEL processes
- show results for two antipatterns
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
9 of 21
23. Approach (2/5): Specify Rules
Context
Start
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
(A,B)
D
(A,B,C)
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
11 of 21
24. Approach (3/5): Process Transform
Context
Motivation
• Process transformation (more abstract and simplified)
(a) from the original BPEL to a simplified BPEL
Related Work
Approach
Experiments
(b) from the simplified BPEL to a generic model
• Goal of this transformation is to ease:
- implementation of the rules
- further analysis of the processes
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
12 of 21
25. Approach (4/5): Process Transform
Context
Motivation
(a)
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
13 of 21
26. Approach (4/5): Process Transform
Context
Motivation
(a)
Related Work
Approach
Experiments
(b)
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
13 of 21
27. Approach (5/5): Detection
Context
Motivation
Related Work
Approach
Experiments
Conclusion
• The implementation of rules
• Applying implemented algorithms on transformed models
• Detection is now semi-automatic
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
14 of 21
29. Experiments (1/5): Objects
travelProcess
3 Web services
7 I/O Var
salesProcess
auctionProcess
3 Web services
6 I/O Var
2 Web services
4 I/O Var
Context
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
15 of 21
31. Experiments (3/5): Results
Context
no detection
N/A
no detection
N/A
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
17 of 21
32. Experiments (3/5): Results
Context
no detection
N/A
no detection
N/A
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
17 of 21
33. Experiments (4/5): Results (cont.)
Context
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
18 of 21
34. Experiments (4/5): Results (cont.)
Context
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
18 of 21
35. Experiments (5/5): Threats to Validity
Context
Motivation
External validity: Possibility to generalize the
results for other large and realistic business processes
Related Work
Approach
Experiments
Construct validity: Different engineers might define
rules differently
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
19 of 21
41. Future work
Context
Motivation
• Perform experiments on other large and complex
business processes
- Limitation: availability of business processes
Related Work
Approach
Experiments
• Automate the approach
• Detect more process antipatterns
• Analyse the processes dynamically
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
21 of 21
44. Experiments (4/5): Results (cont.)
Context
salesProcess
travelProcess
Motivation
Related Work
Approach
Experiments
Conclusion
September 10, 2013
Palma et al. | Detection of Process Antipatterns: An BPEL Perspective
16 of 21