Monitoramento de aplicações baseadas em microsserviços utilizando métricas de desempenho

Abstract

The gains from microservices architecture are increasingly driving companies to migrate their applications to native Cloud Computing platforms. This architecture allows applications to be developed much faster compared to monolithic architectures, since these applications are modularized and decoupled into small services. On monolithic platforms, however, developing new features or even integrating them is much more complex. Microservices have many advantages such as scalability, resiliency, etc. But it also has disadvantages such as remote calls and managing the services themselves. Monitoring resources and analyzing collected metrics in order to identify resource consumption behaviors in a distributed systems environment is neither easy nor trivial. This work proposes a tool called APME, which monitors metrics and detects failures in microservices applications, allows their visualization in time series and is able to indicate, based on a predefined threshold, which microservices were impacted by resource overload. The tool tests were carried out in a reference environment composed of six independent microservices. The tests consisted of load tests in the environment with and without failures. The results show that the tool was able to detect failures and provide system administrators with greater ease in configuring and viewing the collected data, helping them in decision making in the context of failure prevention.