This presentation was created for the SATURN conference series and does not necessarily reflect the positions and views of the Software Engineering Institute.
The presentation provides an overview of and key findings from the
application of the SEI's architectural methods in the definition and assessment
of an architecture for a Geographic Information System (GIS). This application
resulted in the documentation of twenty-two quality attribute scenarios
covering performance, availability, modifiability, security, testability and
usability. Three design iterations were then performed, in accordance with the
Attribute-Driven Design (ADD), producing an architecture, documented in two
architectural views (Module and Component-and-Connector (C&C)).
Thirty-eight distinct architectural design decisions were made; each
contributed to the achievement of one or more quality attribute scenario.
Finally, the GIS architecture was evaluated using the Architecture Trade-Off
Analysis Method (ATAM), resulting in the identification of sixteen sensitivity
points, ten tradeoff points, and thirteen risks, summarised in four risk
themes. Lessons learnt from applying the SEI's architectural methods revealed
that addressing GIS quality attributes systematically at the architectural
stage facilitated an unambiguous record of the rationale, assumptions and
dependencies of the critical technical decisions involved in achieving key
quality drivers. This in turn improved the flexibility, adaptability and
analysability of the architecture. Additionally, the GIS architectural process
proved to be useful for teaching purposes. It is currently used as part of a
postgraduate course in software architecture as an example of a