A provider-side view of web search response time

Abstract

Using a large Web search service as a case study, we highlight the challenges that modern Web services face in understanding and diagnosing the response time experienced by users. We show that search response time (SRT) varies widely over time and also exhibits counterintuitive behavior. It is actually higher during off-peak hours, when the query load is lower, than during peak hours. To resolve this paradox and explain SRT variations in general, we develop an analysis framework that separates systemic variations due to periodic changes in service usage and anomalous variations due to unanticipated events such as failures and denial-of-service attacks. We find that systemic SRT variations are primarily caused by systemic changes in aggregate network characteristics, nature of user queries, and browser types. For instance, one reason for higher SRTs during offpeak hours is that during those hours a greater fraction of queries come from slower, mainly-residential networks. We also develop a technique that, by factoring out the impact of such variations, robustly detects and diagnoses performance anomalies in SRT. Deployment experience shows that our technique detects three times more true (operator-verified) anomalies than existing techniques.

End-Host Driven Troubleshooting Architecture for Software-Defined Networking

“…Next, we sketch up three scenarios: first, assume that Tenant A runs latency-sensitive e-commerce or banking application [6], [7]. Such applications usually have several distributed components (e.g., storage (S), front-end (F)) that require low-latency communication paths to reduce application response times.…”

Detection for Application-layer Denial of Service Attack based on Cluster Analysis

“…Time Series Decomposition [3] decomposes each time series data point into the trend term, periodic term and noise term, and uses different statistical methods for the anomaly detection specific to different components. Each data point in the response latency timing data is decomposed into the trend term (Lt), period term (St), noise term (Nt) through Time Series Decomposition.…”

Identifying Root Causes of Web Performance Degradation Using Changepoint Analysis

“…[1] proposes an approach which identifies root causes of latency in communication paths for distributed systems using statistical techniques. An analysis framework to observe and differentiate systemic and anomalous variations in response times through time series analysis was developed in [9]. A host of research from Borzemski et al highlights the use of statistical methods in web performance monitoring and analysis [2][3][4][5][6].…”