Design And Implementation Of Driver Module For Distributed Database Middleware

With the development of the information technology, People have changed their attitude a lot on data storage and retrieval. People prefer to keep the data in the Internet, such as Social Network Service. And now, more and more data are stored into the Internet instead of personal hard disk. When there are a lot of data, we call them the big data. Today, the development of the database has been grown slower than the rate of the data. We need a distributed database if we want to search the data faster and more stably from the database.As the name implies, distributed database distributes the data into different data storage servers. The middleware will get the data from the right database servers by the rules of data distribution when the application layer needs the data of the databases. And it will reduce the average number of the concurrent, thereby will improve overall the performance of the database. The distributed database middleware strictly follow the MySQL communication protocols and SQL standard. It is also in the middle of the application layer and database clusters. It analyzes, rebuilds and sends the message or query to the corresponding MySQL database instance by capturing the packets which come from client. And then gets the message from the server to send them to the client. The middleware can easily shard the data into the cluster.after that, it can provide read and wirte separation, parallel computing, high availablility and failover. So it has both the advantages of the traditional relational database and distributed database.This paper describes the design and implementation of driver module for distributed database middleware. Drive module contains three parts: data communications, cluster management and the building of execution plan. Data communication mainly uses Adaptive Communication Environment (ACE) to develop. It achieves data communication between middleware and client application, between middleware and the MySQL server, in addition it also can resolve the common MySQL data packets.The function of cluster management are monitoring the state of the database cluster; achieving the read/write splitting of database cluster, load balancing, failover and failback; database fragmentation by the logical partition for database cluster. Read/write splitting and load balancing can effectively alleviate the performance loss caused by high concurrency. Failover and failback improve the reliability of the cluster. Cluster uses dual mutual standby and one-master multi-slave mode which avoids the trouble that a single point of failure causes the entire cluster unavailable.The building of execution plan designs and achieves import data statement LOAD DATA INFILE, sorting query statement SELECT ORDER BY LIMIT and DDL (Data Definition Language). In the process of achieving the execution plan, a lot of optimization and sorting algorithms are used to improve the performance of statements execution.Finally, all the functional requirements are tested. In addition, the performance of data importing statement and sort inquiry statement for the distributed database middleware driver module are tested to ensure that there is no decline of the efficiency and performance in the implementation process. Drive module has already delivered after testing and became an important part of the entire distributed database middleware.