Distributed Tracing in Multi‑Cloud Systems

Abstract

Modern cloud systems rely heavily on tracing tools to monitor application behavior, identify performance issues, and maintain reliability across distributed environments. In this study, we assess the impact of four popular tracing tools Jaeger, Zipkin, OpenTelemetry, and a combined multi-cloud configuration on system performance. The evaluation focuses on three key metrics: CPU usage, memory consumption, and the size of generated trace data. Our results show that Zipkin generally introduces the least overhead, making it a suitable choice for environments where resource efficiency is critical. OpenTelemetry provides a balanced trade-off between observability and system impact. The combined multi-cloud setup, while offering comprehensive trace visibility, results in the highest resource usage. These findings highlight the need for careful selection of tracing tools based on specific deployment requirements, especially in large-scale or latency-sensitive systems. This study offers practical insights for engineers and architects looking to implement effective and efficient observability solutions in real-world cloud infrastructures.