Monitoring of the microservice architecture: Ridango case study

Abstract

Monitoring is an essential part of the software production lifecycle, as it provides feedback about the state and well-being of the observed system, allowing to detect issues and to make informed decisions based on the gathered data. Whether a system is sufficiently monitored depends mostly on its purpose, but often also on its general architecture and structure. In recent years, microservice architecture (MA) has been a common choice for IT systems of all sizes, favored for the various benefits it provides that streamline the development and deployment of modern applications. The MA, however, is significantly more difficult to monitor than most of its predecessors due to the inherent complexity of its distributed and dynamically changing structure, requiring the user to navigate through the various levels of virtualization and correlate the gathered metrics between a service and the rest of the system’s components. Observability tools which have also gained popularity among the software engineering community, aim to solve this problem by combining the monitoring data of metrics, logs, and traces into a single platform while providing real-time analysis of the incoming data with the help of AI models. In this study, we compare three observability tools: New Relic, IBM Instana, and Datadog in their ability to monitor microservices in a self-hosted Kubernetes cluster of a mid-sized IT firm Ridango based on the gathered requirements from a conducted user requirements analysis. Finally, we describe a proposed monitoring solution for the company that is adjusted to its business requirements and the particularities of monitoring the MA.