Research On Structured Access Method Based On MongoDB

With the advent of Web2.0, relational databases face the challenges of massive, high concurrent and high scalable data, while non relational databases can solve these problems. Although MongoDB is a typical and widely used NoSQL database, its performance is not good for handling complicated relational operations. The purpose of the research on MongoDB structural data access is to make the complex relationship operations between documents easy. The MongoDB structural data access is to store the relationship between data in database according to certain data structures and perform relational operations on the data model. Although the official documentation of MonogDB has offered a variety of relational model for one-to-one, one-to-many and tree-structure data, there are a lot of issues exist, such as its one-to-many model does not consider the application scenarios for the data model, and does not put forward many-to-many data model etc.. To solve these problems, this thesis has done the following work:(1) Improve and add structural data storage model of MongoDB. Improve the one-to-many data model in the MonogDB official document and discuss its application scenarios, and add it a data model for handling a large amount of data. For many-to-many relation, propose a data model in which interconnected double array structure is used and it is suitable for a wide range of scenarios. Improve the tree-structure data model in the official document with three kinds of data models.(2) Design the application program interfaces of relational operations on the proposed data model. Implement respectively the SQL-like functions of relational operations based on the proposed data models, and abstract them to a set of interfaces. Application developers can directly call these interfaces to operate MongoDB, which enables the Create, Read, Update and Delete(CRUD) operations upon the proposed data models.(3) Study and provide a solution for transaction processing of MongoDB. Based on the fact that single MongoDB document can maintain the characteristics of transaction, like the relational database, It proposes a simple and widely used transaction processing solution of MongoDB. It gives the flowcharts and implementation of the solution.(4) Verify the model and analyze its performance. Design experiments to test the interfaces by calling these interfaces to operate MongoDB to do the same functionality as MySql does, and compare the time-consuming. By using control variable method, compare and analyze both performance of speed and concurrency for reading and writing.The research work shows that the structural storage data model of MongoDB presented in the thesis can well achieve relational operating on the model; and in the case of small amount of data and low concurrency, it has better performance.