Microservices require introducing observability. Without it, when something breaks, you're never sure where to start looking for the potential root cause. Observability is a way to deduce the state of your application without the need to run a debugger or log to the machines your workload is running on.
A combination of log aggregation, application metrics, monitoring, and distributed tracing is a prerequisite to manage microservices-based architecture. This is especially true once you consider that autoscaling and self-healing may even prevent you from accessing individual services if they start crashing.
