Example of an early sizing, cost and schedule estimate for an application software system

Abstract

Software development has been characterized by severe cost overruns, schedule slippages and an inability to size, cost and determine the development time early in the feasibility and functional design phases when investment decision must be made. Managers want answers to the following questions: Can I do it? How much will it cost? How long will it take? How many people? What's the risk? What's the trade-off? This portion of the paper shows how to size the project in source statements (S/sub s/), how to relate the size to the management parameters [life cycle effort (K) and development time (t/sub d/)] and the state-of-technology (C/sub k/) being applied to the problem through the software equation, S/sub s/ = C/sub k/ K/sup 1/3/ t/sub d//sup 4/3/ . The software equation is then solved using a constraint relationship K = | /spl nabla/D |t/sub d//sup 3/, where |/spl nabla/D| is the magnitude of the difficulty gradient empirically found to be related to system development characteristics measuring the degree of concurrency of major task accomplishment. Monte Carlo simulation is used to generate statistics on variability of the effort and development time. The standard deviations are used to make risk profiles. Finally, having the effort and development time parameters, the Rayleigh/Norden equation is used to generate the manpower and cash flow rate at any point in the life cycle. The results obtained demonstrate that engineering quality quantitative answers to the management questions can be obtained in time for effective management decision making.