@dredmorbius
Not exactly on topic and probably will sound like "old man yells".
This year is the 50th anniversary of the Royce's paper "Managing the development of large software systems". There he describes the problems with the Waterfall approach to the software development. Then gives 5 (five steps) that should improve the results and the whole process.
The frightening thing is that only the first step sounds out of time. But it could be interpreted like "do the capacity planning" - find out a reasonable numbers about the users, discrete queries for processing, how much data stored/retrieved.
And Step 3 is on the spot also. Plan to have a internal prototype within the time-frame of the project in a 1/3 to 1/4 from the whole duration. It will show the where problems should be expected.
The importance of good documentation is also in bold.

These two steps (and the documentation) are still very hard to find currently which is (mildly said) sad. They have always delivered a good results yet they still are rare-to-spot beast in the wild.

I call the "programming" - "the software producing process". I don't want to call it industry because a working industry manages to produce working products. The software producing process still produces a products that are in permanent state of failure (just the extent varyies). Of cource there are exceptions but the exceptions usually stood the trial of time i.e. they are not produced from the current input variables and process.

The paper:
http://www-scf.usc.edu/~csci201/lectures/Lecture11/royce1970.pdf