Multi-master Replication of Enhanced Learning Assistant System in IoT Cluster

In this paper, we discuss the multi-master replication of a Learning Assistance System (LAS). An LAS is an enhanced Learning Management System (LMS) developed in the lab of Wakkanai Hokusei Gakuen University. We consider an LAS cluster where availability of the service is inevitable. Data replication can be used to improve availability. However, the replication process can present several problems, such as consistency, availability, and partition tolerance. There are several ways to address these issues. For example, by developing a consistency-aware protocol, we can increase the data consistency of the application. However, we took a lazy approach to address this issue, in which we utilized a Galera cluster with an isolated sequential replication process. Our experiments show reliable results. The approach taken in our experiment is a combination of active and passive replication, to propagate changes in the background, log every transaction, calculate differentiation of content, and provide a comprehensive solution for replication with an eventually consistent approach. Furthermore, our experiments demonstrate that a multi-master replication system is effective for LAS in an Internet of Things (IoT) cluster. In this paper, we discuss the relevant results that we obtained from the evaluation.