Microservices architecture is widely adopted on various Internet application platforms. However, existing interceptor-based microservice tracing protocols commonly suffer from high host resource utilization and low real-time visualization of full link data. To address these issues, we propose an improved link information collection protocol. By utilizing reflection technology, it dynamically acquires execution parameters and outcomes of business requests, overcoming the limitations of traditional protocols that fail to record complex business request parameters. This approach enables local collection and synthesis of call link data, reducing the resource consumption of the host machine by tracing components. Building upon this protocol, we design a low-overhead microservice tracing system that provides a range of visualization functionalities including service call link tracing, comprehensive display of business flow data across the full link, and statistical analysis of processing durations within call links. After testing and validation, the system is proven that it not only effectively visualizes business flows and service dependencies, but also achieves low overhead and high operational performance.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Microservice Tracing Based on Enhanced Link Information Collection Protocol

  • Yubei Tang,
  • Baili Zhang

摘要

Microservices architecture is widely adopted on various Internet application platforms. However, existing interceptor-based microservice tracing protocols commonly suffer from high host resource utilization and low real-time visualization of full link data. To address these issues, we propose an improved link information collection protocol. By utilizing reflection technology, it dynamically acquires execution parameters and outcomes of business requests, overcoming the limitations of traditional protocols that fail to record complex business request parameters. This approach enables local collection and synthesis of call link data, reducing the resource consumption of the host machine by tracing components. Building upon this protocol, we design a low-overhead microservice tracing system that provides a range of visualization functionalities including service call link tracing, comprehensive display of business flow data across the full link, and statistical analysis of processing durations within call links. After testing and validation, the system is proven that it not only effectively visualizes business flows and service dependencies, but also achieves low overhead and high operational performance.