You have run a load test. When examining the metrics, you see that the virtual users experienced many timeouts and excessive wait times. The system throughput metrics exceeded expected results, even during peak times. Based on your analysis, what conclusion should you draw?

The combination of metrics points directly to a processing bottleneck. High wait times and timeouts indicate that the system cannot handle requests in a timely manner. Simultaneously, high throughput shows the system is accepting and processing a large volume of transactions. This scenario is characteristic of a system whose processing capacity (e.g., CPU, application server logic, database processing) is saturated. It is working at or beyond its maximum capacity, leading to a queue of requests and consequently poor response times for individual virtual users.

A. Network infrastructure should be investigated: A network bottleneck would typically restrict the flow of requests to the server, which would decrease system throughput, contradicting the observation that throughput was high.
C. Virtual users are exhibiting impatient behavior: Virtual users are scripts that operate on pre-configured timeout settings. The timeouts are a result of the system's slow response, not a behavioral choice by the virtual user.
D. New test data is needed: While poor test data can cause issues, the described symptoms (system-wide high throughput and timeouts) are more indicative of a general resource limitation than a specific data problem.

1. ISTQB® Certified Tester Performance Testing Syllabus (2018): Section 3.4, "Performance Test Analysis and Reporting," discusses the analysis of performance test results to identify bottlenecks. It states, "The main goal of performance test analysis is to identify the system bottlenecks (the first component to be saturated by the load)." The scenario describes a saturated processing component. (Source: ISTQB® Official Syllabus).

2. Jain, R. (1991). The Art of Computer Systems Performance Analysis: Techniques for Experimental Design, Measurement, Simulation, and Modeling. Wiley. Chapter 3, "Common Mistakes and How to Avoid Them," and Chapter 33, "Analysis of a Single Queue," explain the fundamental relationship between throughput, response time, and utilization. As a resource (like a CPU) approaches 100% utilization, response times increase dramatically while throughput may hit a plateau or "thrash." The described symptoms align with a system operating at or beyond this saturation point due to insufficient processing speed.

3. Jiang, Z. M., & Hassan, A. E. (2015). A Survey on Load Testing of Large-Scale Software Systems. IEEE Transactions on Software Engineering, 41(11), 1091-1118. This survey discusses common performance metrics. It implicitly supports the conclusion by explaining that response time degradation under high throughput is a primary indicator of a performance bottleneck within the system's processing tiers (application or data layer). (URL: https://ieeexplore.ieee.org/document/7153785, Section III.A. Performance Metrics).