Hands-On Microservices with Kubernetes

Convention over configuration

Sometimes, you don't really need configuration; the program can just make some decisions, document them, and that's that. For example, the name of the directory of the output file could be configurable, but the program can just decide that it's going to output and that's that. The upside of this approach is that it is very predictable: you don't have to think about configuration, and just by reading the code of the program, you know exactly what it does and where everything is supposed to be. The operators have very little work to do. The downside is that if more flexibility is needed, you have no recourse (for example, maybe there isn't enough space on the volume the program is running on).

Note that convention ...