Queue-Based Batch Processing Architecture for Scalable Educational Assessment Systems Using MERN Stack

Abstrak

The increasing volume of student assessment data requires educational information systems that remain responsive during concurrent access and large-scale data import. This study proposes and evaluates a queue-based batch processing architecture for a web-based educational assessment system developed using the MERN stack. The main contribution is the integration of a RESTful API layer, FIFO-oriented asynchronous job queue, and batch segmentation strategy to decouple bulk Excel import from foreground user requests. The system was developed using Extreme Programming to support iterative requirement refinement and continuous testing. Evaluation was designed through functional testing using Cypress, API testing using HTTPie/Thunder Client, load testing using Apache JMeter, and frontend quality assessment using Google PageSpeed Insights. The scalability benchmark compares direct synchronous insertion as a baseline against the proposed asynchronous queue-based batch architecture under multiple concurrency levels, dataset sizes, and batch configurations. The verified PageSpeed results indicate excellent web quality scores, namely Performance 93, Accessibility 96, Best Practices 96, and SEO 100. The JMeter-based metrics, including average response time, 95th percentile response time, throughput, error rate, CPU usage, memory usage, queue waiting time, and job completion time, should be inserted from the exported test logs before final submission. The proposed architecture is expected to improve responsiveness, prevent server overload during bulk import, and provide a more reliable foundation for scalable educational assessment management.