Partition Storage In The Data Register Center

Under the background of the information age, from the physical world to the virtual world network, data are rapidly growing. The value contained in the data has become increasingly diverse, economic value is also increasingly being taken seriously, in a certain sense to become an important asset. But how to solve the problem of information silos, to contact them together, and how to manage the use of these data of the physical world and the virtual world bring users and technology developers a lot of trouble and challenges. At present, although software development techniques are diverse, but did not have good methodology for data. Data Oriented Architecture is in this context has been raised.In different systems, the way of data storage is different. Whatever the structure, Data storage and system performance are closely related. Data is the core of the Data Oriented Architecture, and the driving force of the system data. In massive data, the performance of the Data Registry Center as the core of the data management system affects the whole system. The traditional single point type can not play an advantage in massive data, distributed data registration center becomes an inevitable choice.This paper, under the original results of the“data-oriented architecture”,study the characteristics of the data in the Data Registry Center,the tends of the metadata in the background of big data, and the distributed storage of metadata. The main content contains:(1) What is“all things are data”. Analysis the difference of the “all things are data” with the traditional data in the Data Register Center.(2) Study the characteristics of metadata and its role in the management of big data. Based on the characteristics of the current metadata and the development of the status quo, combined with the characteristics of the Data registration center, a data model suitable for Data Register Center management of generalized data is established.(3) Study the distributed storage of metadata. In depth analysis of the characteristics of the traditional database and distributed database, as well as the characteristics of the current distributed database, it is compared as a metadata distributed storage platform.(4) Study the optimization of data storage organization structure and designed the structure. Data manipulation contains insert data, read data. Different table structure, different sizes of the insertion and query will affect the performance. According to the requirements on the number of inserts, statistical characteristics of data access,propose a three segment storage structure. And it is used in the construction of the Data Register Center.The innovations of this paper are mainly in the following aspects:(1) The distributed design of data registration center is carried out. Enhance the applicability of the Data Register Center, and improve the ability of data Register center to deal with massive data.(2) A model of three segment storage and query is established. On the basis of No SQL distributed storage, according to the data insertion and the cost of pure access data, and the data access problem, set up the first area. According to the characteristics of data access, a new algorithm is designed, which is based on the short time effect and the frequency of access. And applied to metadata distributed storage. According to the algorithm, the second area and third area are set up. the paper establishes a model of the three-section type storage and query.