Design and Implementation of a Distributed Call‑Chain Tracing System for Microservices

In microservice architectures a single request often traverses multiple services, modules, and machines, making it difficult to determine the exact call chain, locate performance bottlenecks, and reproduce issues.

To address this, a distributed call‑chain tracing system is introduced, which assigns a globally unique TraceID to each request and propagates it across service boundaries.

The system records the order and parent‑child relationships of calls by generating a SpanID for each invocation and linking it to its ParentSpanID, enabling the reconstruction of the full topology.

An independent, non‑intrusive Agent is deployed on each host to automatically create and forward TraceID, ParentSpanID and SpanID, typically by injecting them into HTTP headers or other transport protocols.

The Agents send collected data to a central collector, which aggregates the information and visualizes the complete call graph, allowing developers to pinpoint slow services and understand system behavior.

The article also mentions existing open‑source tracing solutions such as SkyWalking, Zipkin, and Pinpoint, and notes additional concerns like sampling rates, ID generation algorithms, and UI design.