Development and Deployment of Microservice-Based Architecture System for Efficient Application Loading in Cloud Environments
Cloud-based applications require architectures that can maintain efficient application loading and performance as workload increases. This study developed and experimentally evaluated three functionally equivalent e-commerce implementations: Monolithic, Baseline Microservices, and Proposed Microservices architectures. An experimental research methodology was adopted, with the same business logic, database, and workload scenarios maintained across the implementations. The Proposed architecture integrated Docker, Kubernetes on Amazon Elastic Kubernetes Service (EKS), horizontal pod autoscaling, load balancing, Redis caching, and CloudFront delivery. Controlled workloads were generated using k6, and performance was assessed using startup time, deployment availability, response time, throughput, autoscaling, resource utilisation, inter-service latency, full user-journey performance, cache effectiveness, and frontend loading metrics. Results showed that the Monolithic architecture had the shortest startup time (1.33 s), while the Proposed architecture required 25.36 s. Under 2,000 virtual users, however, the Proposed architecture achieved an average response time of 33.47 ms and throughput of 1,915.78 req/s, compared with 1,951.56 ms and 670.13 req/s for the Monolithic architecture. The Proposed architecture also recorded 0.00% deployment failure, scaled from 2 to 15 replicas, and reduced cache latency by 30.12% (p = 0.0001). Its frontend FCP (First Contentful Paint), LCP (Largest Contentful Paint) and TTI (Time to Interactive) were each 617 ms. The findings demonstrate that integrated cloud-native mechanisms can substantially improve application performance and scalability under increasing workloads.