| With the rapid development of computer technology and widely application,the amount of global average daily data increased exponentially,the traditional single-machine mode system could not effectively analyze the large-scale data.The emergence of distributed computing framework Hadoop has greatly reduced the development cost of distributed system,and the demand for distributed system development has increased rapidly.In this paper,based on the theory of layered architecture model,reference to the design of the modular and componentization principles,combined with the characteristics of the distributed system development,design and implements the application system development framework IMSAA based on Hadoop platform.The IMSAA framework divides the system into five levels: Data Import Layer,Data Management Layer,Data Service Layer,Data Application Layer and Data Access Layer.Through the division of this layered framework,the process of distributed system development is simpler,and the division of labor is clear at all levels,and the coupling degree is greatly reduced.The main contents of this paper include: firstly,analyzing the development background of the era of big data,investigating the current situation of relevant theories and technologies at home and abroad;Secondly,the function demand and non-functional demand analysis of IMSAA framework are analyzed.Then,based on the design idea of layered architecture and plug-in architecture,the IMSAA framework was designed in general and the six modules of the framework were designed and implemented in detail.Finally,an IMSAA framework is used to design a mining technology and economic evaluation system,and verify the feasibility of the IMSAA framework through system test.The result of system test show that IMSAA framework make software architecture more reasonable,improve the speed and efficiency of the distributed application system development,effectively reduce the coupling between modules,are proposed for distributed system design and development of new solutions. |