Big Data Management And Analysis Oriented Cloud Computing Model Research

Developments of big data oriented massively distributed network application architecture model and its application software efficient development and deployment method is one of the major challenges big data management and application faces.The traditional parallel relational database,data warehouse,retrieval index,application modeling and software development technology,has been unable to meet the actual requirement of the development and deployment of big data application.At present,the correlative technique of big data organization and management commonly exists low level formalization of technical model,support big data partition management university is not strong,not support big data organization pattern based on heterogeneous data integration,lack of support big data semantic analysis application and other issues.Achievement of big data oriented massively distributed application architecture and efficient development and deployment of application software is rare.This study is derived from top-level design and engineering project of national business information service platform of certain department,which is undertaken by National Engineering Laboratory for Modeling and Emulation in E-Government.Around the seven application problem on big data of the project,developing related research on big data partition management and massively distributed application architecture and efficient development and deployment of application software,and final implemented as three infrastructure platform and successfully deploy the information sevice platform of certain department on it and launch business application.The detailed research achievements are as follows:1?On big data organization and management research:Proposing the formalized data partition approach TSEI-PS that is made up of the big data partitioning rules based on the activity scenes and the identification of entity instances,and the activity scenes oriented fragment allocating rules;thus establish big data partition management model with universal operability.Afterwards,providing differentiated big data information resource organization model based on big data partition management model after extended its trace property and item functional which decribes differentiated information,and combining with big data resource retrieval problem and big data resource retrieval operation,big data resource inverted retrieval mode is established with highly efficient query and retrieval and its higher efficiency than traversal retrieval mode and hierarchical retrieval mode has been proved as well.Thereout,forming the data organization and management platform,which supports big data partition management,and supports difference shielding and efficient query and analysis.2?On massively distributed application architecture:Establishing the model of interactive computing category that is constituted by protocol algebra,interactive computing bus lattice algebra and multi-structural state relational algebra and morphisms among these objects and indentify morphisms,and conformed to definition of directed ordinary category,through introducing directed ordinary category.On this basis,proposing Model of Interactive Computing(MIC)which serves as the theoretical model of researching interactive computing,through defines the morphisms of model of interactive computing category as micro-component,and thus set of the same type micro-components is forms distributed object system component which is defined by DOBPS.Meanwhile designing a distributed architecture load dispersing mechanism which serves for big data application,by taking architecture model of Cougaar and DOBPS as lessons.And then big data management and analysis oriented cloud computing information service platform is established by using this mechanism.3?On efficient development and deployment of application software:Defining Problem of Distributed Application on Big Data(PDADB)formally,and giving general depict of solving form of PDADB that consists of domain locating operator,domain dividing operator,sub-domain computing operator,solution aggregating operator,and composite sub-domain computing operator.Forming development and deployment formalization method of massively distributed software base on MIC and big data management and analysis oriented,through mapping all kinds of operators into micro-components of MIC,and applying MIC to problem solving of PDADB.