Optimizing Cloud-Native Services with SAGA: A Service Affinity Graph-Based Approach

Modern software architectures are characterized by their cloud-native, modular, and microservice-based designs. While these systems are known for their efficiency, they also face complex challenges in service optimization, especially in maintaining end-to-end quality of service across dynamically distributed services. This paper introduces a novel approach using the concept of Service Affinity to address this challenge. The proposed method, termed Service Affinity Graph-based Approach, employs a graph-based model to model the interactions among microservices. It formulates the service placement as a minimum-weight k-cut problem and utilizes an approximation algorithm for service clustering. This approach is realized through a conceptual framework that takes into account a wide range of optimization objectives, ranging from enhancing application performance and enforcing data privacy to optimizing operational costs. In addition to presenting the SAGA framework in details, this paper conducts an in-depth empirical evaluation using a prototype deployed on a Kubernetes cluster. The results demonstrate a mean latency improvement of 23.40 %, validating the effectiveness of our approach. Finally, the paper comprehensively discusses various aspects of the proposed methods, including their impli-cations, challenges, and benefits, providing a thorough analysis of the approach's impact.