Would you decide to buy a car without knowing its mileage, maximum speed, cost, maintenance cost (service), etc.? And after buying, will you not keep track of the expenses you incur on the car to judge whether you got value for money?
A software project in many respects is similar. It is like car. When a client buys a software, the purpose , design (model of car), performance (speed of car), cost, total cost of ownership (service for a car, fuel efficiency) etc all need to be considered. Ofcourse, this is all in addition to the fundamental need - a defect free product. Similarly for a manufacturer of software there has to be considerations around aspects like productivity.
Does all of above make it sound complex..? Lets decode it
Before even we discuss software metrics, lets talk about measurements in software. There are, at first level, just 4 basic measurements.
1) Effort
2) Defects
3) Size
4) Schedule
These 4 measurements all have a single unit associated with it. Like Person hours for effort or a Function Point for size. All are independant units of measurement. It sometimes amazes me that collecting and tracking these 4 in a projects is considered a huge task!!
It is possible to compute innumerous Metrics from these 4 measures. Many efficiency and effectiveness related measurements are possible using these 4 measures. There have been additional basic measures that have been used, but that does not add the complexity of the efficiency and effectiveness, but it shows that the industry is maturing. 100 years back, doctors never knew that there is a relationship between blood pressure and heart attack. Now there is a relationship that has been established. Similarly software metrics have developed. Today, we are researching about relationships bewteen Coupling between Objects, Cyclomatic complexity etc on defect rate, productivity and cost of maintenance etc. Software metrics are maturing..
More about metrics soon . . .
No comments:
Post a Comment
Help us improve this blog site by leaving your comments