Research On Data Registration Centre For Metadata Storage In DOA

With the rapid development of information construction, different kind of data are created, serving people’s daily life at every moment. The spread of the Internet makes data flow in the network, and scatter in different data storage systems and devices. In order to realize polymerization at client side, G/S model is created. G/S model adopts "request-polymerization-service" mechanism to organize and manage the service information procedure, and realize the data aggregation and service polymerization. G/S model solved the problem of data aggregation, but how to manage various kinds of data, G/S model does not give an effective solution. In order to acquire data faster with higher quality, this article put forward Data-Oriented Architecture (DOA) with data as the core and identifier as the main line, which through metadata management realize data integrated management.In order to manage different kinds of data, first of all, metadata is needed, which is the emphasis of DOA. Due to the diversification, massive volume, and isomerism of data, different industry has different metadata standard. But the industry feature of these metadata, makes them cannot be the standard of metadata. DOA require data registration center can extract metadata fast, how to storage metadata is the problem to be solved, and it is the bottleneck of metadata management under DOA, so this article study how to solve the problem of metadata fast obtain under data registration center.This article described G/S model and DOA at first, analyzed their structure and the function of data aggregation and monument, and the reason of metadata storage. Through the study of metadata standard and development framework, lay a solid foundation of metadata storage standard.The main work of this article include:(1) Research metadata storage strategy under data registration center.DOA unified management of data by using metadata, so the speed of getting metadata will decide the performance of DOA. The key of getting metadata fleetly is how to store metadata. Research methods in the analysis of the file system, LDAP directories storage, a variety of relational database storage and memory data storage approach based on the advantages and disadvantages, evaluated whether they can fulfill the requirement. Based on those mentioned, put forward a model of metadata storage for getting metadata fleetly.(2) Study algorithm of metadata storage.Based on the problem of memory size and hit rate of metadata, studied related algorithm for memory metadata loading. Studied a cache replacement algorithm based on cost function, in order to save frequently accessed metadata dynamically under the data registration center. Based on these mentioned, further studied the algorithm of choosing high hit rate metadata tables for the memory database. By these two algorithms, realized the high hit rate metadata stored in memory database.The innovative points and achievements of this article are:(1) This paper put forward a metadata fast obtain solution in data registration center under DOA.This metadata storage solution, according to the requirement of unified data management by metadata under DOA, using relational database storage capacity, memory database cache dynamic memory access efficiency and role of cost-based cache replacement algorithm stored in the cache dynamic may be frequently accessed recently metadata to improve acquisition speed and accumulation metadata is access frequency; further cost function based cache selection algorithm to select the most frequently accessed cache metadata tables stored in memory database, complete memory database metadata store. On this basis, the design of the data under the registry DOA metadata storage structure model, and the model was designed based on the data under the registry DOA metadata storage system.(2) This paper put forward a cache replacement algorithm for replacing the table of the minimum access value in data registration center.According to analyze several existed common cache replacement algorithm, from the perspective of function, use cache data stable size, frequency of access, access time and the time that CPU get the cache data stable. Based on the mentioned, designed a cache replacement algorithm on the cost function and realized that frequently accessed data table is stored in the cache.(3) This paper put forward a selection algorithm that high access rate of metadata was stored in memory database in the data registration center.Based on the analysis of cache replacement algorithm, from the perspective of function, use cache data stable size, frequency of access, access time and density. Based on the mentioned, designed the algorithm of screening metadata tables on the basis of cost function. By these two algorithms, realized the high hit rate metadata stored in memory database.