Engineering a Reliable Data Infrastructure for a Mileage Application: Replication Strategies and Predictive Data Engineering Study

This thesis presents the design and implementation of a SQL Server database system based on a distributed approach, database used for managing the lifecycle of motorcycle components in a racing environment. This thesis has been conducted over a project proposed by Aprilia Racing company, part of Piaggio Group. The project analyzed the previous existing system, redesigned the database schema to accommodate new requirements, and implemented a replication strategy to ensure data availability in a distributed environment. The database schema was upgraded and designed to track the mileage, maintenance operations, run configurations, and anomalies of each component, recording, in this way, the entirety of their lifecycle. The replication system was set up using SQL Server's built-in features, allowing for sychronization between a central publisher and multiple subscribers. The implementation details, the faced challenges, and the solutions adopted during the project are discussed in detail. This thesis also explores the potential for leveraging the collected data through data prediction using machine learning techniques. The results demonstrate significant improvements in data integrity and user experience, highlighting the effectiveness of the new design and implementation. The Machine Learning models showed promising results in predicting component anomalies based on input data features, laying a good foundation for future works in this direction.