Methods and Models
Measurement of the size of the product to be delivered and problems to be solved drive the predictive models and must be comparable with benchmark data. Historically Source Lines of Code (SLoC) were measured, this has become impracticable as modern applications are vast and employ varied technologies. As a result Functional Size Measures (FSM), often referred to as Function Point Analysis (FPA) evolved to measure size. The three most commonly used methods, which we use, are COSMIC FFP, MkII FPA and IFPUG FPA.
Measurement of Functional Size
Functional Size Measures are technology independent, systematic and objective methods for analysing, estimating and measuring the functionality of a software system. A typical large company that has been developing applications for thirty years can easily have accumulated in excess of 100,000 FP so the required effort involved can appear daunting. To resolve this the following methods have evolved.
Three methods of sizing
These techniques conform to the ISO Standard for the Functional Size Measurement of Software ISO/IEC 14143:1998 and each has approval as an ISO/IEC standard.
COSMIC FFP (Full Function Points) – ISO/IEC 19761
Used in a wider range of problem domains, such as real-time or infrastructure development, as well as the business/MIS domain. The method caters for software system types; avionics, telecoms, production systems and so on. It has has been specifically designed to fit with the complex multi-layered architectures and approaches to modern software development. COSMIC FFP demonstrates an advance in generality, simplicity and accuracy.
MkII FPA– ISO/IEC 20968
A method designed to address the structured approaches in software design. The MKII method is equally useful in the area of object oriented design, fitting well with the use case notation. It is the preferred method of the UK government as it is easily applied to large systems.
IFPUG FPA – ISO/IEC 20926
The original method of sizing, it is currently at version 4. This method is still the most widely used and works well in the business/MIS domain. It is applicable to object oriented developments.
|
