Puppet modules are getting better and better, they are more reusable, they better fit doing just what they are supposed to do, and they offer stable and reliant interfaces to the management of different applications.

They are therefore getting nearer to that status where they can be considered shared libraries that can be used by different users to compose the configurations needed in their environments.

Here is where many people are struggling in order to achieve a sane organization of resources, good patterns to group them, and better approaches to a dynamic, reproducible, structured, and maybe reusable management of complete stacks of applications.

At this higher abstraction layer module, people are experimenting with different practices. Some have become common, like the roles and profiles pattern, but still few are engineered with the vision to be reusable and eventually allow single components to be composed freely to fit different topologies.

In this chapter, we focused on this topic, reviewing why we need higher abstraction modules and why it would be great to have them reusable and flexible.

There's still much to do on this topic and I think much will be done in future years.

In the next chapter, we move to completely new topics. We will explore the practical challenges we face when we have to introduce Puppet, either on a new or an existing infrastructure.