Automated Model Management for Microservices: A CI/CD Approach

Abstract

The rapid advancement of microservices architecture in software and internet applications has introduced significant challenges in managing distributed systems, particularly in maintaining operational efficiency and real-time responsiveness. This paper presents an innovative automated encapsulation and deployment framework tailored to address the frequent updates required by data-driven models in dynamic microservices environments. By integrating Docker containerization with Continuous Integration/Continuous Deployment (CI/CD) practices, the framework streamlines model management processes, enhancing both adaptability and operational efficiency. Leveraging modern tools such as Kubernetes and GitLab, we developed and tested a prototype system that facilitates rapid updates and robust management of intelligent models, ensuring their responsiveness to evolving data patterns. Empirical evaluations demonstrated substantial reductions in model deployment times and improved responsiveness to business requirements, while also identifying critical bottlenecks in traditional packaging and deployment methods. Our findings underscore significant improvements in operational efficiency and reduced manual intervention for model updates, with further optimizations to minimize deployment times and resource consumption. This research contributes a scalable and efficient solution for managing the lifecycle of models in microservices environments, addressing key challenges such as service granularity, dependency management, and rapid deployment needs, thereby enhancing overall system performance and reliability.