Instrumenting Observability in a Decentralized Microservice Architecture

Abstract

Software systems have increased in complexity over time, and with this increased complexity has come an increased need to keep these systems organized and functioning efficiently. Observability is closely attached to ensuring this correct and effective system function. Without system monitoring, it is difficult to pinpoint when errors occur and correct them at their sources. Monitoring systems also helps to understand a system from the outside by allowing developers to ask questions about the system’s state and function without needing to know the details of what comprises the system’s internal behavior. While there are existing solutions for observability frameworks, these solutions do not target microservice architectures, which are used more and more with expansive code bases, such as those likely to be employed in an industry environment. They also require extensive configuration to be fully integrated with a pre-existing system. As such, the challenge lies primarily in adapting observability solutions to a decentralized, microservice architecture found in an industry setting. The existing solutions also come with advantages and disadvantages for different situations, so they are often incomplete in addressing an entire system’s needs. The integrated system created here satisfies our system’s requirements of a consolidated observability platform while also enabling future customizations, thereby allowing problems to be identified more quickly and proactively.