The path towards observability in a microservices architecture

• Autor(a) principal: Ferreira, Sónia Cristina Teixeira
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.22/24322
• topic: Microservices; Observability; Monitoring; Tracing; Centralized logging; Domínio/Área Científica::Engenharia e Tecnologia

The availability and performance of client applications is a huge factor for the success of organizations. The digital transformation of manufacturing processes is making use of modern development paradigms, such as microservice architectural styles, a growing trend for the last years. Like any other architectural style, it has its costs and benefits. Microservices use a distributed system to improve modularity but, distribution is a cost. One other cost is the operational complexity. While being able to deploy small independent services it also has increased the demands on managing and monitoring those services. Observability becomes essential to ensure stable services, allowing us to understand and explain the inner workings of our applications, even in unpredictable situations. The aim of this dissertation is to design a road map to develop and implement observability and monitoring for distributed systems. This includes exploring the current tools available that allow monitoring, using the three pillars of observability (logs, metrics, and tracing) and evaluate which one or which ones are the most efficient and effective for the microservices architecture being studied, since identifying an appropriate solution, out of all available options, actively exploring and evaluate it, takes time that usually developers don’t have. Several approaches were tested and compared using Chaos Testing, measuring the time it took from the moment a fault was injected into the system until an alert was generated. Additionally, for each approach, the quality of the information obtained in logs, traces, and metrics was evaluated. Although the results obtained did not provide sufficient evidence to conclude that any ap proach was more efficient than the others, the knowledge gained during the development of this dissertation has enabled us to define the necessary steps to make Target App into an observable system.