OPTIMALISASI SKALABILITAS (STUDI KASUS: E- COMMERCE)

As e-commerce applications continue to grow rapidly, a major challenge is handling the increasing number of requests and data volume, especially during peak transaction periods. Based on e-commerce behavior data from Kaggle (April 2020), Add to Cart activity peaked at 9:00 AM with 6,658 actions per hour, indicating a significant traffic surge in a short time. E-commerce platforms must efficiently manage user interactions and transactions while supporting high scalability. To address this, the Event Driven Architecture (EDA) offers an effective solution, where system changes are represented by events communicated asynchronously through a message broker such as RabbitMQ. This research developed an e-commerce prototype implementing RabbitMQ as the message broker and Docker for managing event-based services. The study aims to evaluate the effectiveness of this architecture in handling high request volumes during the add-to-cart process. Load testing using K6 was conducted to assess system performance based on metrics such as latency, throughput, and scalability.
The test results show that the system using the Event-Driven Design (EDD) approach is able to maintain stable performance even as the number of users increases. In the test scenario with 50 Virtual Users over 30 seconds, the EDD-based system successfully handled up to 14,859 requests with an average response time of 49.96 ms and a throughput of 493.82 requests per second. In contrast, the system using the Remote Procedure Call (RPC) approach under the same conditions was only able to handle 169 requests, with an average response time of 56.90 ms and a throughput of 4.44 requests per second. This difference indicates that the implementation of RabbitMQ as a message broker, supported by Docker, can significantly enhance the scalability and performance of the system.
Keywords: RabbitMQ, Docker Event Driven Design