An extensible lightweight framework for distributed telemetry of microservices

Microservice architectures have become the standard for developing scalable distributed systems that offer significant benefits in managing the integration and evolution of complex applications. However, they face challenges in effectively diagnosing and resolving performance and reliability issues. Traditional centralized telemetry models and cloud-based monitoring platforms often require complex or costly configurations and are not optimized for RESTful microservices. In fact, although the OpenAPI Specification (OAS) has become a key standard for describing microservice APIs, existing telemetry tools do not leverage this information to enhance service analysis and diagnostics. This paper introduces a lightweight and distributed approach to telemetry that uses OAS-based API information, offering an automated, configuration-free system that enables developers and operations teams to perform root cause analysis more efficiently. Moreover, we propose a plugin system to incorporate intelligent behavior into the telemetry system, such as an adaptive proactive alert mechanism when response-time anomalies are detected. By incorporating this extensibility mechanism, the framework paves the way to address issues such as energy consumption and performance, allowing the system to dynamically adjust its monitoring activities to optimize resource usage and minimize the carbon footprint of microservice deployment and execution. This adaptability reduces operational overhead and supports sustainable computing practices. To validate our approach, we present a proof-of-concept in the form of a ready-to-use package for the NodeJS ecosystem, demonstrating that this distributed telemetry model can operate with minimal impact on system performance and resource usage, proving its effectiveness to support more robust and sustainable IT systems. • Decentralized Telemetry : Lightweight, distributed system for microservices using OAS. • Efficient Overhead : Cuts resource use vs. centralized models, with minimal impact. • OAS-Based Monitoring : Automates monitoring with context-aware insights without setup. • Plugin Extensibility : Supports plugins to attach intelligence (e.g. spot anomalies).