Architecture Optimization Design And Realization Of National Land Confirmation Registration System

With the rapid development of Internet technology,all professions and trades have moved into the age of big data,and the size of data that people produce and need to handle is more and more.These data are only the carrier of information,and have the characteristics of polymorphism,heterogeneity,mass and variety.Recently,mining the potential value of these massive data for the further decision-making of the enterprise has become the focus of all trade's attention.At present,the system has been put online and operated steadily,and accumulated a great deal of data on the work of confirming and registering certificates of land contract and management rights.Including the user information and data information of 27 provinces(autonomous regions),344 cities(municipalities)and 2859 counties(districts)across the country.Each report data consists of information from the county and the province.Each county(district)submits no less than 42 items of the information data,and provincial and municipal users need to statistics of not less than 38 items of information.The data increased by 134176 records per month,increased by 536704 records per quarter.The report date generated in a year can reach to 2146816 records,in addition to the basic information data submitted by users at all levels.With the increase of data size and the continuous expansion of subsequent functions,the performance of concurrent processing and large-scale data analysis of the existing system will sharply decline,and the existing system cannot meet the requirement of system expansion.Based on the above problems,this thesis studies the technologies related to massive data processing.Based on the researches on the key technologies of data warehousing,distributed storage and computing,combined with the operational requirements of the national land registration system,we optimized the massive data processing architecture of existing system.The main work and research results are as follows:1)According to the business requirements of the Ministry of Agriculture,data warehouse based on data analysis is designed.Through the data warehouse dimension model,combined with the data warehouse ETL technology,the process of extracting,cleaning,converting,and reloading from the national land confirmation registration system's business database to the data warehouse has been realized.It provides a first-hand decision support plan for massive data analysis in the Ministry of Agriculture,which reduces the risk of decision-making ineffectiveness and decision-making interference caused by old data.2)A distributed cluster system is built by using the distributed file system of Hadoop,and the data transmission between data warehouse and distributed file system is realized by using Sqoop technology.Combined the distributed storage of files in cheap hardware devices,the proposed distributed cluster system solves the massive data storage problems arising in the existing system with the continuous expansion of functions,and lays the foundation for subsequent distributed computing models.3)Based on the research on Hadoop's distributed computing model MapReduce,combined with the data analysis requirements of the national land ownership registration and software and hardware installation,an efficient distributed computing model is designed to optimize the parallel processing capability of the system.The MapReduce's mapper and reducer processes are designed in detail,and correlation matching is realized in the distributed server.The final result will be obtained from each server data.This design reduces the burden caused by frequent network communication,compared to the original system with the advantages of parallel computing,asynchronous processing,greatly improves the speed of data processing.