Measuring Round-Trip Response Latencies Under Asymmetric Routing

2025-05-20

AI Summary

This paper addresses the challenge of passively measuring end-to-end response latency under transport-layer encryption (e.g., TLS/QUIC), where application-layer headers and client instrumentation are unavailable. We propose PIRATE, a passive latency estimation algorithm that relies solely on observable client→server traffic. Its core innovation is the first use of causally linked request-pair time differences—without client-side instrumentation or plaintext headers—to accurately proxy application-layer round-trip latency, inherently supporting asymmetric routing. The method integrates causal request-pair detection, passive temporal modeling, and DSR-aware load-balancing coordination. Evaluated on real-world web services, PIRATE achieves ≤1% estimation error for client-side latency. When deployed at Layer-4 load balancers, it reduces tail latency by 37%, significantly enhancing bottleneck identification, adaptive scheduling, and attack mitigation capabilities.

Technology Category

• Reasoning under Uncertainty: Causality
• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Causal Learning

Application Category

• Responsible Web: Measurement, analysis, and circumvention of Web censorship
• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Latency is a key indicator of Internet service performance. Continuously tracking the latency of client requests enables service operators to quickly identify bottlenecks, perform adaptive resource allocation or routing, and mitigate attacks. Passively measuring the response latency at intermediate vantage points is attractive since it provides insight into the experience of real clients without requiring client instrumentation or incurring probing overheads. This paper presents PIRATE, a passive approach to measure response latencies when only the client-to-server traffic is visible, even when transport headers are encrypted. PIRATE estimates the time gap between causal pairs - two requests such that the response to the first triggered the second - as a proxy for the client-side response latency. Our experiments with a realistic web application show that PIRATE can estimate the response latencies measured at the client application layer to within 1 percent. A PIRATE-enhanced layer-4 load balancer (with DSR) cuts tail latencies by 37 percent.

Problem

• Measure response latencies under asymmetric routing conditions
• Passively estimate client-side latency without client instrumentation
• Improve load balancing by reducing tail latencies significantly

Innovation

• Passively measures response latencies without client instrumentation
• Estimates client-side latency using causal request pairs
• Enhances load balancers to significantly reduce tail latencies