Traditional logging, such as writing string-formatted log messages to files, is the most common root cause for poor performance. This chapter has already described the issues and advisable solutions for them.

The biggest reason for poor performance is the extensive string object creation and resulting memory consumption. High memory consumption, in general, represents a major performance issue. This is not only caused by logging but by high memory rates in caching, memory leaks, or extensive object creation.

Since the JVM manages the garbage collection of memory, these high memory rates result in garbage collector runs, trying to free unused memory. The garbage collection utilizes the CPU. The situation is not resolved by a single collection run, what results in subsequent GC executions and thus high CPU usage. This happens if not sufficient memory can be freed either because of memory leaks or a high workload with high consumption. Even if the system doesn't crash with OutOfMemoryError, the CPU usage can effectively stall the application.

Garbage collection logs, heap dumps, and measurements can help with investigating these issues. JMX tools provide insights about the memory distribution and potential hot spots.

If business logic is implemented in a lean, straightforward way using short-lived objects, memory issues are far less likely.