Analisando o efeito de cargas de trabalho no ambiente virtual de aprendizagem Moodle com o JMeter na AWS

Abstract

Shortly after the emergence of the internet in the 1990s, advancements in new technologies for human benefit began. Among these, Virtual Learning Environments (VLEs) emerged, aiming to teach students through the internet. However, these environments require significant resources to deliver satisfactory results. For this reason, a widely discussed alternative has been the use of cloud computing due to its flexibility, resource management, and low cost. Nevertheless, cloud computing alone does not address all issues related to efficient resource usage and maintaining proper functionality. Therefore, performance evaluations are necessary to establish limits and enable future improvements. In this context, this study will analyze the effects of workload on the Moodle Virtual Learning Environment (VLE), hosted on the Amazon Web Services (AWS) cloud platform, using the Apache JMeter tool for performance testing. The main objective is to evaluate Moodle’s capacity under different load scenarios, identifying bottlenecks and determining the optimal configuration to handle variable demands. Three instances (M7i-flex, M6a, and M5n) were created on the EC2 service, with properties aligned with the study’s objectives, where the Moodle servers will be hosted. Next, the necessary performance tests were selected, including stress tests, load tests, and smoke tests, along with the planning of test scenarios simulating activities such as login, navigation, and forum interactions. The workload was divided into five tests (ranging from 150 to 3000 threads), executed by Apache JMeter, while performance was monitored using the CloudWatch service, with selected metrics such as CPU usage, latency, and error rates. The results revealed that the M7i-flex server performed the best, handling up to 500 simultaneous threads without critical failures. Higher loads led to performance bottlenecks and even temporary unavailability. Meanwhile, the M6a and M5n servers showed lower resilience, with frequent failures and greater sensitivity to overloads. Latency increased proportionally with the load, particularly in tests exceeding 1000 threads. The study demonstrated that the choice of AWSinstance significantly impacts Moodle’s performance. Proper configurations can improve scalability and availability, which are essential during periods of high academic demand.