A Reputation-based Microservices Trust Model Using Similarity Domains

The Microservices Architecture (MSA) has become one of the most prevalent distributed systems architectures in recent years. It has been widely adopted by industry to build large-scale systems because of its advantages in terms of scalability, elasticity, service independence, reusability, and agility. As with any other distributed system, one of the challenges that MSA faces is establishing trust between microservices. The management of trust within MSA is of great importance, as the concept of trust is involved in microservices communication, and the operation of an MSA system is highly reliant on communication between these fine-grained microservices. Previous research has introduced some trust models that can be used for trust management. Many researchers have suggested the use of MSA as a means to build open systems, in which microservices from any providers may join or leave the system without restriction. Open microservices systems need to cope with the unpredictable open-world setting and the arbitrary behaviour of microservices. In order to support the operation of open microservices systems, a trust model should also be able to manage trust in an open environment. However, current trust management solutions are designed for closed systems. Therefore, it is important to analyse the limitations that current microservices trust models have in an open environment, identify the qualities that trust models should have to support open microservices systems, and design and build a trust model that can work with open microservices systems using these qualities. This thesis addresses the critical challenge of trust management in open microservice systems. The study recognises the inadequacy of existing models in the context of open systems and the characteristics of open systems and proposes eight essential qualities that trust models should exhibit to support the openness of microservices systems. To address these needs, a reputation-based microservice trust model using similarity domains is proposed. This model leverages Natural Language Processing (NLP) and Machine Learning (ML) to evaluate functional similarities between microservices based on standardised OpenAPI documentation and incorporates it into the domain management of microservice systems. Additionally, a comprehensive trust bootstrapping process that collectively enhances the system's ability to manage trust without compromising on performance metrics, a reputation-based trust score computing scheme that uses both clients' ratings and Quality of Service (QoS), and a probabilistic trust balancing scheme are adopted. This allows the trust model to cope with the characteristics of open systems and fills the existing gap in trust management for open systems. The proposed trust model is evaluated in a simulated open microservice system. The model demonstrates strong performance in managing microservices and their trust in open systems. This thesis contributes to the field by offering a scalable and secure trust management framework that, unlike previous work, is adaptable to the dynamic nature of open microservice systems.