Multitasking does not work

By Queena Lee-ChuaPhilippine Daily InquirerFirst Posted 17:15:00 06/06/2010

Filed Under: Education, Science & Technology,

Booch and Architectural Reviews

Why ?

• Stakeholders (paymasters) will want to see that the significant design decisions properly address their quality requirements and the business value creation needs.
• It is valuable and educational to compare the existing design decisions with architectural alternatives
• It is necessary to (identify and qualify) triage and fix what is broken.

See SEI ATAM method.

How ?

• Identify the forces that act on the system under developement
• "Grok" the system's essential architecture (essential versus accidental)
• Generate scenarios that exercise these forces against the architecture
• Throw the architecture against the scenarios - see what distorts.
• Ask hard questions
• Don't go down rabbit holes
• Follow the smells and press on the system's tender spots
• Keep it civil
  

Notes:

Review should be cheap, and periodic.
The architecture is not always readily available. A shared language is required, shared views and understanding. At least a list of significant design decisions is required.
People colour the reality they present based on where they wish to go.
A review can become emotive, empassioned, this can be a good sign. Expect grit, drama, avarice as we are human.
Assure that architectural diagrams/models are real - review naked code with architecture.