The categorization is a bit artificial - I just could not press 200+ pages of patterns into a single part of the book. Thus there are 4 parts, each covering 4 or 5 patterns.

The first part is monolith and results of its division along each of the axes of the structural diagrams that I use.

The second part covers more complex systems - those that add a layer to a system of services (but it happened that they may be applied to other kinds of systems, e.g. a proxy may run on top of a monolith).

The third part conjoins patterns that don't feature a single line along either abstractness or subdomain axes of their diagrams - which means that they look fragmented like brick walls.

And there remain a special group of patterns which don't fit into of the above categories - instead each of them is found to implement a component of a system - a mainly monolithic module (Plugins, Hexagonal Architecture) or a middleware (Microkernel, Mesh).

Thus, the categorization is based on the structural diagrams of the metapatterns - in the same way as each metapattern is a category of patterns based on their structural diagrams.