Why architectures are evaluated, the three methods, and the nine ATAM steps with the outputs an evaluation is supposed to produce.
| Concept | Definition |
|---|---|
| Risk | A decision or gap that endangers a quality goal — e.g. failover never tested under load |
| Non-risk | A decision checked and found sound, with the reason recorded |
| Sensitivity point | A property where a change strongly affects one quality attribute |
| Trade-off point | A property that is a sensitivity point for two or more attributes at once |
| Method | Focus | Use when |
|---|---|---|
| SAAM | Scenario-based modifiability and functionality; came first | The question is how costly a set of planned changes will be |
| ATAM | All quality attributes and their interactions; generalized SAAM | A finished architecture has competing quality goals |
| ARID | Active design review of an intermediate, unfinished design | A partial design must be validated by the teams that will use it |