Design and Empirical Evaluation of a Zero-Trust Cloud Database Pipeline for High-Concurrency Web and Mobile Applications

Abstract

This paper presents the design, implementation, and empirical evaluation of a novel Zero-Trust (ZT) database pipeline architected specifically for cloud-native, high-concurrency web and mobile applications. Traditional perimeter-centric security models fail under the dynamic, distributed nature of modern cloud environments.¹ The proposed architecture enforces "never trust, always verify" principles, eliminating implicit trust zones and mandating strict, granular authentication and authorization for every data access request, regardless of origin.² Key features include mutual TLS (mTLS) for all service-to-database communication, dynamic authorization using Policy-as-Code (e.g., OPA), and data-in-use protection via confidential computing techniques. The empirical evaluation, conducted under simulated peak load conditions ($>10,000$ concurrent connections), assessed the pipeline's security efficacy, latency impact, and scalability compared to a conventional cloud security group/VPC-based pipeline. The findings demonstrate that the ZT model significantly enhances the security posture by mitigating common cloud vulnerabilities (e.g., credential exposure, insider threats) with only a marginal, acceptable overhead on transaction latency and overall throughput, thus establishing a scalable blueprint for securing sensitive data in demanding application environments.