Infrastructure Engineering For Data-Driven Software: Building Robust And Scalable Systems

Abstract

In the era of data-intensive computing, the performance and resilience of software systems are increasingly dependent on underlying infrastructure design. This study investigates how infrastructure engineering influences the robustness and scalability of data-driven software systems by evaluating five architectural models: Monolithic VM, Microservices VM, Containerized Microservices, Kubernetes Autoscaling, and Edge Hybrid. Using a combination of real-world case studies, controlled performance benchmarks, and statistical analyses, we assess metrics such as uptime, response time, throughput, recovery time, and machine learning inference accuracy. Results show that Kubernetes and Edge Hybrid architectures consistently outperform traditional models, demonstrating superior fault tolerance, self-healing capability, and elasticity under load. ANOVA and regression analyses confirm statistically significant differences across infrastructure types, especially in recovery metrics and predictive performance. Visualizations further highlight the relationship between infrastructure complexity and reduced system downtime. These findings reinforce the strategic value of infrastructure engineering in supporting high-availability, low-latency, and scalable applications. The study offers actionable insights and a reproducible framework for practitioners aiming to align infrastructure design with the demands of modern, data-driven software ecosystems.