Vorhersagen mehr als Schätzen - Die Softwarelieferung mit Simulationen prognostizieren

Authors	published	Link
Ralf Westphal	2022	Find the book here

Summary

In his book Ralf Westphal takes on a pain that software development organisations all over the planet face: "When will this be ready?".

The simple question is very simple, yet often stirrs up a lot of inter-personal problems.

Ralf Westphal creates a journey for us on which he explains an idea how to organize that in a way that will make all stakeholders happy, including the customers, the product owners and the developers.

More Details

• Organizational Structure
  • The developers are responsible for creating good software.
  • The product owner is responsible for the creation of value in a software development environment. He can speak to the customer and get a picture of the valuable tasks that need to be performed. He is also responsible for the ganularity of the tasks.
• For Forecasts:
  • The developers are responsible for giving the product owner an impression of solution complexity and the hurdles involved, when asked.
    Deadlines must not be given to the developers as they are not responsible for time.
  • The product owner is responsible for collecting the statistical data and creating a forecast.
    The product owners job is it to create forecasts and, if necessary, navigate or reduce the amount of remaining work as well as to discuss with the customers.
• At first the author explores some different ideas how to approach the forecast from different angles, finally concluding that the Monte Carlo Simulation is the best that can possibly be done.
• He then uses the concept and shows what times need to be gathered to get forecasts going.

There are two important durations that need tracking:

• Cycle Time
  The cycle time is the time that you actually are working on something from start to finish, including wait times.
• Lead Time
  The lead time is the time from the order of the client to the delivery of same order. So it is the cycle time including the time before the work can actually start and the time it takes to ship the thing to the customer.

Using those two concepts you can create ideas of the throughput through your production system. Using throughput-per-day measures and the Monte Carlo Simulation you can create ideas of duration (your forecast) for either a set of tasks, or, like shown in chapter 22, can predict throughput in a certain amount of time. The latter gives you the ability to limit planning to a possibly working amount.