Distributed tracing in systems using Kafka is challenging due to Kafka's decoupled producers and consumers, which lack direct transactions to trace. However, monitoring Kafka clusters is possible with tools like OpenTelemetry, which offers two main approaches: zero-code and manual instrumentation. Zero-code instrumentation involves using an agent to automatically provide telemetry signals without changing the source code, offering quick setup but limited customization and visibility. Manual instrumentation, on the other hand, allows for more detailed and customized insights by adding dependencies and code, although it requires more effort. The blog demonstrates these concepts using a sample application with Kafka components, highlighting the differences in visibility between the two methods and encouraging exploration of OpenTelemetry's capabilities for enhanced application insights.