240 likes | 353 Views
Domain-specific Languages for Service-oriented architectures: an explorative study. Agenda. Problem Background Our Approach Study Details Study Results Future Work Conclusion. Problem. PROBLEM. PROBLEM. Domain Experts. High level of abstraction, domain-specific concepts.
E N D
Domain-specific LanguagesforService-oriented architectures:an explorative study
Agenda • Problem • Background • Our Approach • Study Details • Study Results • Future Work • Conclusion
PROBLEM Domain Experts High level of abstraction, domain-specificconcepts Low level of abstraction, technical-, technology-, platform-specificimplementations IT Developers
Domain-Specific Languages (DSL) • tailored for narrow domain • e.g. SQL • multiple levels of abstractions • business vs. technical experts
Model-Driven Software Development (MDSD) • Functionality vs. Technology • Different Levels of Abstractions • Business Level: • no technical knowledge • Technology not important • IT Level: • multiple technologies • collaboration between technologies • changing permanently (e.g. versions)
Our Approach • Design decisions/Trade-offs • MDSD-based DSLs • basic SOA concerns • e.g. controlflow, informationflow • extensional SOA concerns • e.g. transactions, human interactions • non SOA concerns • e.g. pageflow of Web applications
Claims • systematic development approach • multiple levels of abstractions • domain experts • IT experts • extension/integration points
Study Results • Claims: • systematic development approach • process-driven and non-process driven SOA • separation into high- and low-level DSLs • domain and IT experts • enhance understandability and readability • reduce complexity
Study Results • Design Decisions: • DSL syntaxlanguage model • High-level models low-level models • multiple models one model
Study Results • Trade-Offs: • Separation into high- and low-levels • redundancy in languages • inconsistencies • overlapping concerns • Detailed separation • complex model merging • less understandable and readable DSLs
FUTURE WORK • more experiments • process-driven SOA • non-process-driven SOA • non SOA • more design decisions/trade-offs • more evaluation
CONCLUSION • Problem • domain vs. IT experts • MDSD-based DSLs • Study Details & Results • Design decisions/Trade-offs • Three experiments • Future Work