架构质量属性

1 1 Addressing Quality Requirements in GIS Architectures Ibrahim Habli GIS Services Division Khatib & Alami (CEC) Lebanon Tim Kelly Department of Computer Science University of York United Kingdom TS14.1 SIM and Quality From Pharaohs to Geoinformatics, FIG Working Week 2005 and GSDI-8, Cairo, Egypt April 16-21, 2005 2 High Quality GIS Attention is often paid to GIS functionalities However, quality aspects are insufficiently addressed. A GIS would be: �� Ineffective Ineffective if its processing misses deadlines �� Unreliable Unreliable if it is not available when it should be �� Unusable Unusable if it is difficult to understand Hence, high quality GIS systems depend on qualities, such as � Efficiency � Reliability � Usability � Security 3 Problems with GIS Quality Requirements Complexity and large volume of geographic information Often not systematically captured & documented Common to find ambiguous statements such as: �� ““System shall be portableSystem shall be portable”” �� ““System shall be highly securedSystem shall be highly secured”” �� ““GIS operations shall be efficientGIS operations shall be efficient”” Hence, no feasible means to assess whether the system has met its quality requirements or not 4 Solution: Architectural Framework Based on two architectural techniques from the S Software E Engineering I Institute (SEI): (SEI): � Quality Attribute Scenarios � Attribute Driven Design Method (ADD) Quality Attribute Scenarios & Use Cases Conceptual Architecture 5 Quality Attribute Scenarios SEI quality attribute scenarios consist of 6 yardsticks: 1. Source of stimulus 2. Stimulus 3. Environment 4. Artifact 5. Response 6. Response measure Source: GIS administrator Stimulus: The administrator requests to convert the data format of a portion of the GIS data Artefact: GIS system (Data source) Environment: Runtime Response: The required format is converted with no data inconsistency Response Measure: Number of elements affected/ programming effort/ data loss 6 Attribute Driven Design Method Attribute Driven Design Method (ADD) is a recursive approach to software architecture design based on the quality attributes the software needs to achieve 1. Choose one design element 2. Choose the architectural drivers i.e. quality requirements 3. Choose architectural patterns 4. Assign functionality to each of the design elements (resulting from the decomposition) 5. Verify that the decomposition has addressed the selected architectural drivers Quality Attribute Scenarios & Use Cases Conceptual Architecture 2 7 Achieving GIS Quality Requirements using ADD Quality requirement scenarios addressed: � Performance � Editing a Geographic Feature � Retrieve Data � Modifiability � Change GIS Data Format � Add GIS Component � Interface GIS with an External Software System 8 First Level of Decomposition <> GIS <> User-Interface <> Applications <> Data <> OperatingSystem <> Communication • Scenarios addressed at this stage: • Add GIS Component • Change GIS Data Format • Editing a Geographic Feature • Retrieve Data • Architectural Decisions: • Maintain semantic coherence • Published interfaces • Client-Server • Minimize clients & servers interaction 9 Second Level of Decomposition: Applications Subsystem Decomposition • Scenarios addressed at this stage: • Mediator design pattern • Use geographic information standards • Separation unit operation • Architectural Decisions: • Interface GIS with an External Software System • Change GIS Data Format • Editing a Geographic Feature • Retrieve Data <> User- Interface <> Applicat ions <> Data <> OperatingSystem <> Communication 10 Quality Attribute Scenarios: � Editing a Geographic Feature � Retrieve Data � Change GIS Data Format � Add GIS Component � Interface GIS with an External Software System <> DataFacade <> AttributeData <> GeographicData <> FileBasedData IGeographicData IAttributeData IFileBasedData IDataFacade <> DataSecurity IDataSecurity <> ApplicationSecuri ty <> Services <> Mediator <> Real-TimeServices IServices IApplicationSecurity IMediator IReal-TimeServices <> Application Facade IApplicationFacade <> GUI IGUI <> ThickGUI IThickGUI <> ThinGUI IThinGUI <> ExternalInterface IExternalIn terface <> SWInterface ISWInterface <> HWInterface IHWInterface <> UIAdapter IUIAdapter User-Interface Subsystem Data Subsystem Applications Subsystem Architectural Decisions: 1. Maintain semantic coherence 2. Published interfaces 3. Client-Server style 4. Minimize clients & servers interaction 5. Data accessor design pattern 6. Façade design pattern 7. Concurrency 8. Caching 9. Increase hardware resources 10. Mediator design pattern 11. Geographic information standards 12. Separation unit operation 13. Generalization style 14. Adapter (or wrapper) design 11 Evaluation: GIS Quality Attribute Scenarios Understandability: scenarios unambiguously define factors controlling the achievement of quality attributes Precision: response and response measure offer specific means for assessing GIS architectures Traceability: decomposing each quality attribute into scenarios enables traceability of how an attribute is addressed during the architectural design and evaluation 12 Evaluation: GIS Architecture Design Attribute Driven Design Method: • Simplifies architectural design process • Systematic consideration of quality attributes • Mapping between quality attribute scenarios & architectural decisions Design Documentation: • Well organized architectural documentation • Record of architectural design decisions applied, resultant architectural views and underlying design rationale 3 13 Q&A