Enhancing Distributed System Performance with an Intelligent Multi-Agent Microservices Architecture
Résumé
In a rapidly evolving digital landscape, distributed systems are becoming increasingly complex, making it essential to adopt architectures that can adapt in real-time. Business process optimization now depends on intelligent, dynamic, and modular systems. To address this need, we propose a hybrid architecture called multi-agent microservices (MAMS), which merges the principles of microservices with those of multi-agent systems (MAS). Each microservice is represented by an autonomous agent that can discover, compose, and collaborate with other services through REST interfaces while adapting in a decentralized manner to contextual changes. The uniqueness of our approach lies in integrating deep reinforcement learning (DRL) into the decision-making process. This technique enables agents to learn how to flexibly select optimal service combinations based on quality of service metrics, such as availability, response time, and reliability. Unlike traditional methods, our system ensures continuous adaptation to load fluctuations or service failures. The main contributions of this work include intelligent and autonomous distributed orchestration, decision-making driven by adaptive learning, and significant improvements in flexibility, resilience, and efficiency of services in complex environments. A case study on travel planning demonstrates the effectiveness of our solution, which surpasses monolithic architectures in personalization, user satisfaction, and operational efficiency.
Citer ce document
Accès au document
Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter
Voir l'article sur le site de la revueAuteur(s)
Statistiques
Consultations : 1
Téléchargements : 0