Research And Application Of A Big Data Storage Model

Storing and processing big data using ordinary relational database comes across some problems. Some data storage systems have turned to NoSQL for solutions. NoSQL databases give up the powerful relational operations and may not support all the applications that were once based on relational data store.This paper intends to propose a new big data storage model based on relational data store. The solution not only meets most requirements of big data storage and processing, but also supports most of the relational operations. As a result, transplantation of the applications based on relational data store can be easily carried out.In order to achieve the goal, this paper has studied some typical big data storage model in depth, including several NoSQL storage model and some distributed relational storage architectures. With the key difficulties of big data storage using relational data store in mind, the paper introduced the thought of object-oriented and designed a3-dimension data relational model which is able to store data with complicated structure. Data is partitioned by classes in the paper based on the3dimension relational model. This paper uses a serious of optimization measures to make parallel execution available and eventually forms complete big data storage and processing model. The model has been tested in actual cluster for improvement and perfection.This model originates from and has been applied with remarkable achievements in the National Free Pre-pregnancy Health Check Project, which is part of the National "Eleven Five" Science and Technology Support Plan. The storage model solves the scalability problems of relational data store and improves data query efficiency. It can be used as a data storage optimization solution for relational data store for solving big data storing and processing problems, especially where it is unavailable for existing data and programs to transplant to NoSQL database.