Research On Distributed Processing Of Incremental Data In Network-based Social-security Audit

In recent years, with the development of the Internet and distributed computing, the performance of massive data operation has improved. In many application areas,it's necessary to extract incremental data from the massive data to meet the demands of a business, especially it is required to extract fix time incremental data from continually updated source data. Therefore, how to quickly identify the incremental data for extraction, storage and transmission, will gradually become the focus of attention of foreign scholars.This thesis is on the background of the social security network audit.It extracts the data as incremental source data for the audit predict system.It makes ' post-audit' turned into ' concurrent audit' to keep abreast of changes in the social security data, and found that the audit in doubt and forecast laws of datas based on the data changes. Incremental data recognition and extraction technology is a key technology for the entire system.This thesis describes the development of distributed systems, load balancing strategies and the incremental data recognition technology research situation at home and abroad. A method of identification the incremental data has been provided ,which is based on distributed processing. This paper basically realized incremental data processing system ,which included four modules:data partitioning,data distribution,incremental data identification and task reallocation, and data recovery. A platform has been built based on performance analysis and system design .In this paper a distributed approach combined with multi-threading technology has been used, and load balancing has been carried out according to various sub-nodes allocation of tasks.In the end, using a province audit data as experimental data,simulation experiments have been done on the platform which is builded on dawning cluster of the Audit Office .By the performance tests ,it shows that the platform can basically meet the functional requirements of network audit operations .It completed an efficient identification of incremental data ,and the timeliness of the audit predict system has been enhanced.