Building small applications and building large systems are very different. Small software projects of a 100 function points can be built by a few developers and a few other personnel. These are low in risk and usually delivered on time. Large systems in the 10,000 function point size range require many skilled specialists such as business analysts, architects, and data base analysts who may not be needed for small projects. These are extremely risk and almost never on time or within budget. Worse, quality control is often poor so post-release bugs are troublesome for users.
This paper shows typical results for three size ranges: 100, 1,000, and 10,000 function points. The data comes from benchmark studies by the author and colleagues at Namcook Analytics LLC. Our Software Risk Master (SRM) tool was used to provide the quantitative data in the paper and in the following summary table:
|The Impact of Size on Software Project Costs and Schedules|
|(Assumes IT financial application and Java programming language)|
|Size in LOC|
|Team staff size|
|$ per function point|
The differences in function point size leads to very different kinds of development practices and to very different productivity rates at the low end compared to the high end. For example for some large systems finding and fixing bugs and creating paper documents cost more than the code itself.
Successful results for large systems requires early sizing and estimating using tools such as Software Risk Master (SRM) and careful progress and cost tracking using tools such as the Automated Project Office (APO).
Vice President and CTO, Namcook Analytics LLC
Copyright 2017 by Capers Jones. All Rights Reserved.