Abstract

The increasing use of multi-cloud infrastructures has created new difficulties in tracking and diagnosing performance issues across different cloud providers. Congestion in inter-domain traffic is a common issue, but traditional traceroute tools fall short of diagnosing inter-domain congestion for lack of perspective and bounded protocols. This paper introduces a diagnostic framework based on enhanced traceroute variants aimed at accurate bottleneck detection and localization in multi-cloud interconnects. The methodology applies a combination of active probing, delay correlation analysis, and next-hop verification to target specific delay hotspots and asymmetric routing patterns. Adaptive cloud network policies and configurations are addressed by a combination of Paris traceroute, TCP-based probing, and hybrid ICMP/TCP diagnostics. With the goal of multi-regional validation, experiments were conducted across AWS, Azure, and Google Cloud Platform from distributed vantage points. These findings suggest that cloud exchange points and ISP-level handoff zones often host traffic and congestion-based anomalies during peak traffic. Augmented asymmetry of routing and round-trip latency differences enrich bottleneck detection. The framework not only increases transparency for opaque multi-cloud architectures, but also enables cloud providers to make better informed decisions on interconnection and routing strategies. The specific method suggested here surpasses standard traceroute techniques in accuracy, detail, and level of confidence in the diagnosis, thereby enabling efficient real-time troubleshooting of cloud networks. The research highlights the need of specialized diagnostics in multi-cloud architectures while paving the way for further developments focusing on machine-learning-enabled automated root cause analysis.