What is the impact of selecting various phase staffing shapes? When should I deviate from the default Rayleigh pattern, and how is my estimate affected?
We have found that the default Rayleigh pattern is the staffing pattern that best matches the application of effort to the work to be performed but due to staffing constraints or different software management styles, you may decide that another staffing pattern fits your organization or project better.
In general, the various staffing shapes can be described as follows:
- Front Load Rayleigh peaks at approximately 40% the phase
- Medium Front Load Rayleigh peaks midway through the phase
- Medium Rear Load Rayleigh peaks at approximately 75% of the phase
- Rear Load Rayleigh (Phase 3 only) peaks at the end of the phase
- Level Load maintains a constant staffing level over the entire phase
- Exponential (Phase 4 only) gives the most rapid drop off of staffing from the end of Phase 3.
- Stair Step (Phase 4 only) begins at about half of the staffing level from the end of Phase 3 and stair steps down.
- Straight Line: (Phase 4 only) represents a straight-line decrease in staffing from the end of Phase 3.
- Rayleigh (Phase 4 only) is a natural tailing off/continuation of any Phase 3 Rayleigh curve. Not valid if the selected Phase 3 staffing shape is Level Load.
Level loading is generally seen in the development of very small systems. In larger systems, early application of people often represents wasted effort. The Default Rayleigh shape determines the staffing shape based on the size of the application. QSM has found that small projects (less than 18,000 lines of code) tend to have a Front Load Rayleigh profile while larger systems (greater than 100,000 lines of code) typically reach peak staffing towards the end of phase 3. For very small systems (3 to 6 months in duration with a peak staff of 1-3 persons), a level load profile is more appropriate.
The estimated total effort for each of the staffing patterns for the first three phases will remain constant regardless of the pattern selected. Phase 4 effort will vary, depending on the shape of phase 3. This is because in phase 4, manpower tails off from the final Phase 3 staffing level and is therefore a function of the number of people on board at the end of phase 3 and the length of phase 4. Projects that peak at the end of Phase 3 will have higher phase 4 effort. Projects that peak earlier will have lower phase 4 effort.
If you are using default QSM overlaps for the phases, you can expect to see some differences in the phase overlap when you change the staffing pattern. This occurs because SLIM attempts to produce the smoothest aggregate staffing pattern when moving from one phase to the next.
For example, the overlap percentage between Phases 2 and 3 will be 0% if Phase 3 is Level Loaded. Otherwise the overlap ratio will range from 33% to 40% as the phase 2 staffing shape changes from Front Load Rayleigh to Medium Front Load Rayleigh.