Design And Implementation Of A Distributed Database Based On MySQL Database

In recent years, with the advances in information technology, the Internet industry is flouring, constantly changing people’s ways of working and living. Due to the continuous expansion and accumulation of data and the rapid growth of Internet services, the increasing requirements of data storage are difficult to meet. Traditional databases cannot cope with massive data processing, and the emerging NoSQL technologies can not completely replace the traditional relational database systems. Therefore, building a relational database which is capable of handling the storage of massive data has become the target of this study. A SEDA-based, parallel processing and distributed database is the direction of this research. Based on the subject of "The pilot study of Intelligent transportation" from the Chinese Academy, this thesis design and implement a massive data oriented distributed relational database-to solve the fast data retrieval of massive bayonet data, with many conditions, providing a technical solution with high performance, high availability and scale-out properties.This thesis discusses the key technologies needed to build a Distributed Database System based on MySQL, like SEDA concurrency model, load balancing, data partitioning, database availability, distributed coordination services, etc. This thesis completed the detail architecture of the system, detailing the main functions and key points on implementation of proxy&computing cluster, storage cluster, metadata cluster, and admin tools.The main contributions of the paper is to design and implement a distributed database cluster:for a distributed table, use the method of horizontal fragmentation; for a data fragment, use the reading and writing separation tech of MySQL; use an external middleware to provide a unified interface. Based on these, the distribution of storage and computing of the database cluster is done. The specific related works are:Design and implement a separate read and write MySQL Cluster with high availability and load balancing.Design and implement a database middleware based on the SEDA architecture, providing a unified service, and hiding the complex, lower cluster architecture.After the testament, the proposed distributed database cluster shows its capability of handling mass data. The querying on a single table is much faster than a standalone MySQL instance. And the high availability, scalability, and load balancing, resource isolation are achieved, and tested.