Service Call Chain Analysis for Microservice Systems

Abstract

Industrial practitioners widely adopt the microservice architecture to build applications. An application with microservice architecture can be composed of a set of individual services. Although microservice can improve the scalability of a system by isolating services, the complexity and difficulty of defect detection and analysis grow. High verification cost, a long feedback cycle, and high communication cost pose challenges to the maintenance of microservice systems. To address the problem, we propose a call chain tracing and analysis approach designed for the microservice architecture. To evaluate its effectiveness, we implement our approach as a plugin, namely Cam, to monitor and analyze exceptions by tracing call chains. Currently, Cam is packaged as a maven plugin for applications using Spring Cloud, which is an opensource microservice framework for Java programs. We experiment it with a microservice system to demonstrate its availability. The result shows that Cam can help software engineers understand the workflow of a service call and locate potential defects.