Abstract
The widespread adoption of microservices has transformed the design, development, and operations of web systems. While microservices offer greater flexibility and scalability than monolithic architectures, they also introduce complexities in monitoring and maintaining system performance. Ensuring resilience amid growing service interdependencies requires robust observability strategies and proactive performance evaluations. This thesis assesses how integrating observability and load testing tools can enhance the scalability and performance of microservices-based applications, specifically focusing on four widely adopted open-source tools: Prometheus, Grafana, Jaeger, and k6. A systematic literature review was conducted to critically evaluate research on microservices monitoring, tracing, visualization, and load testing. These tools were selected due to their prevalence in industry and their complementary functionalities: Prometheus collects real-time metrics, Grafana provides versatile visualization capabilities, Jaeger enables distributed tracing of service interactions, and k6 simulates realistic traffic loads to assess performance under stress. The evaluation criteria included integration ease, metrics collection capabilities, scalability support, visualization usability, operational overhead, and licensing flexibility. The thesis identifies clear research gaps, particularly the need for enhanced automation, tighter integration between observability and system orchestration, and the application of machine learning for predictive scaling and anomaly detection. The findings discover that no single tool offers a comprehensive observability solution. Instead, significant benefits arise from integrating these technologies into a cohesive observability pipeline. Combining Prometheus and Grafana provides real-time system performance insights, Jaeger offers detailed visibility into request flows and service interactions, and k6 actively validates system resilience under simulated load conditions. Together, these tools help quickly locate performance issues, effectively analyze their causes, and check if the system can handle growth. Their open-source nature further encourages widespread adoption, customization, and operational integration.