A Proposal for a Secure and Observable Microservices Architecture for AI-Powered E-Commerce Applications


Creative Commons License

Polat O., Baydağ N., Söğüt E.

International British Congress on Interdisciplinary Scientific Research & Practices-VI , London, İngiltere, 18 - 21 Haziran 2026, ss.256-261, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: London
  • Basıldığı Ülke: İngiltere
  • Sayfa Sayıları: ss.256-261
  • Açık Arşiv Koleksiyonu: AVESİS Açık Erişim Koleksiyonu
  • Gazi Üniversitesi Adresli: Evet

Özet

This study proposes a microservices-based architecture that addresses the requirements of security, scalability, and observability in AI-powered e-commerce applications. In this study, an e-commerce system decoupled from the traditional monolithic structure is designed as a set of components comprising a web application, an AI service layer, an SQL Server database, a Redis cache and state management system, a Prometheus metrics collection infrastructure, and a Grafana monitoring dashboard. At the core of the proposed architecture lies a service approach based on the Model Context Protocol (MCP), which carries context between large language models (LLMs) and external data sources. Through this layer, sentiment analysis, question-answering, and content processing functions are executed in an isolated manner. The system is further supported by prompt injection prevention, data masking, DDoS/DoS mitigation, rate limiting, vulnerability scanner detection, bot blocking, an XSS sanitization process, and custom security metrics. The resulting architecture offers an integrated solution that applies a defense-in-depth approach against the new attack surfaces created by AI services. The study presents a system design based on engineering practice, and more comprehensive experimental validation with quantitative performance results is planned for the future.