Geek Answers Handbook

How many maintenance developers are required per 1000 lines of code

We have a really huge project with 20-30 modules, but basically this is done. It is under maintenance (mostly bug fixes and rarely new features). I am trying to come up with a number of developers that will be needed to support the product.

Is there a good way to measure this number?

These are mainly C # -based projects based on WinForm (a combination of .net 1.1 and 2.0) using vb6 applications.

+6
maintenance
source share
6 answers

This will depend entirely on the quality of the code, the frequency of changes and the level of testing.

For example, a system with thousands of lines of code, but very rare changes and a complete library of modulation / integration tests, will probably require fewer developers so that a small system changes frequently and is not tested.

Another important factor is the experience of the developers involved, not only in general, but also, in particular, their understanding of a particular project.

After all, these are very complex statistics to evaluate, and you are probably best off looking at the workload of the developers who are currently in the project and, if necessary, slowly moving people to or from the project.

+19
source share

It depends more on the number and complexity of errors that need to be fixed, and not on the size of the project. At the beginning:

  • How many mistakes do you get each week that need to be fixed?
  • How complicated are these errors?
  • How high quality is the code base? It makes a huge difference in the perceived "difficulty" of error correction.
  • How much experience do the programmers have in the language / framework?
  • How much experience do programmers have with this particular application?
+2
source share

I think it depends on many variables: the quality of the developers you hire, their familiarity with the code, the credibility of the code and the language used for beginners. I don’t think there will be any equation that will work.

+1
source share

I do not think there is a good way to choose this. This will depend on several factors:

  • Code Base Status
  • Level of experience / talent engineers who work on it
  • Timeline of the expected fix / function error

You may just want to start with a small team, see how the work works, and add new members later if necessary.

+1
source share

If you used a problem tracking system for the duration of the project, you can run a report with the average number of issues per month and the average time for correction, which will then give you a historical service requirement.

You could extrapolate this into the future.

+1
source share

There is a command line tool, sloccount , which implements the COCOMO model . It is available on the website and on Debian based systems via apt-get .

0
source share

More articles:


All Articles