Distributed systems such as cloud-native architecture pose some unique challenges. The sheer number of different services working at any given time makes it very inconvenient to investigate how well the components perform.
In monolithic systems, logging and performance monitoring are usually enough. With a distributed system, even logging requires a design choice. Different components produce different log formats. Those logs have to be stored somewhere. Keeping them together with a service that delivers them will make it challenging to get the big picture in an outage case. Besides, since microservices may be short-lived, you will want to decouple the life cycle of logs from the life cycle of a service that provides them or a machine that hosts the service.
In Chapter 13, Designing Microservices, we described how a unified logging layer helps manage the logs. But logs only show what happens at a given point in the system. To see the picture...
