Systems Analysis and Design
Life cycle and methodologies
Phases are problem recognition, feasibility study, requirements analysis, system design, development, testing, implementation, and review and maintenance. Methods include waterfall, prototyping, RAD, JAD, spiral and agile. Conversion strategies are direct (big bang), parallel, phased and pilot; parallel is safest but costliest.
Fact finding and modelling
Techniques are interviews, questionnaires, observation, document review and sampling. Modelling tools include data flow diagrams (process, data flow, data store, external entity), the context diagram, entity relationship diagrams, the data dictionary, decision tables and trees, structured English, flowcharts and use case diagrams.
Feasibility and design
Feasibility is technical, economic (cost-benefit, payback period, net present value, return on investment), operational, schedule and legal or regulatory. Design covers inputs, outputs, files and databases, processes and user interface, together with controls and security. Logical design is technology-independent; physical design names actual hardware and software.
Trap: a context diagram contains exactly one process, no data stores and only external entities and flows; a level-0 diagram explodes that single process into major sub-processes. DFDs show data movement, not sequence of control, which is the flowchart's job.
Worked example: for a payroll system the external entities are Employee and Bank; the processes might be Validate Timesheet, Calculate Pay and Produce Payslip; and the data stores are Employee Master and Payroll Transaction. Rule of thumb: before submitting any DFD, check that the four element types are consistently named, numbered and balanced between parent and child diagrams.