Tracing in complex service mesh systems is essential for understanding the flow and timing of requests that span multiple internal entities, which cannot be effectively captured by traditional logging plugins. Kong's Zipkin plugin addresses these needs by collecting structured timing data compatible with HTTP/HTTPS and stream traffic. This data is organized into spans and traces, facilitating the debugging of timing-related issues and providing insights into multi-service environments. A basic setup using Docker and Docker-compose involves deploying a Zipkin server, along with HTTP and TCP Echo services, and a db-less Kong instance, configured through a declarative file. The process of tracing involves using Kong to proxy requests and then analyzing the resulting traces in the Zipkin UI, which displays detailed timing information for each request. While HTTP traffic supports tracing headers like x-b3-traceid, TCP traffic does not, highlighting the plugin's capability to manage different traffic types within a service infrastructure.