Building an Observable System Based on OpenTelemetry
Xi-Zhen Wang, Chien‐Chao Tseng · 2024
In traditional monitoring systems, the health status of services is typically assessed through metrics and logs. However, as the scope of monitoring expands from a monolithic architecture to a distributed system composed of multiple services or more granular microservices, traditional monitoring systems struggle to effectively pinpoint specific errors and their root causes. We propose an observability framework based on OpenTelemetry to generate trace data that captures contextual information and event correlations of requests thereby extending the existing monitoring system into an observability system. The study leverages a Hook to implement instrumentation in a PHP environment, ensuring minimal intrusiveness. Additionally, it traces intra-service function calls and inter-service requests to provide a thorough understanding of the execution path of requests.