Research On The Design Of Learning Resource Distribution Model Based On Hadoop

With the development of computer technology, network and multimedia technology,the learning ideology and forms of learners change continuously. In this situation, networklearning becomes one important approach for learners, which can fit into the needs oflifelong study and open study of learners. Therefore, the organizing and providing oflearning resource on network forms into the core of the whole learning process. In order tosatisfy every learner, every organization develops its own resource database. However, withthe development of each subject, the enrichment of learning content and the updating oflearning resource forms, the organizing of the resource becomes the core. Traditionalresource database takes the method of upgrading the hardware to fit the storage of vastlearning resource. But with the boom of learning resource, the bottleneck of hardware makeslots of equipment retired, which decreases the economical efficiency and takes too muchinvestment, but it cannot solve the problem radically.The use of cloud computing in resource database can provide new solutions for thestorage of vast learning resource. Through the analysis of the document literature, distributedcluster storage can be found to be the best solution. After analyzing the current developmentsituation of learning resource and the research of resource database and cloud computingapplication at home and abroad, this paper designs one distribution model for vast learningresource based on the cloud computing platform of Hadoop combining with the keytechnology and storage framework of Hadoop. This distribution model can provide higherstability, expansibility and storage efficiency, which is constructed on the low-pricecomputer clusters by using HDFS as files storage and HBase as metadata storage.The focused research of this paper is to design the learning resource distribution modelbased on the research on the key technology of Hadoop and the features of vast learningresource, which aims at the problems of high storage memory occupation, vast diskfragmentation and high pressure on center node when operating vast small files. This modelowns a four-layer construction model with the core of Hadoop, which designs merge storagestrategy for small files and data structure for metadata to fit into the inner relationshipbetween vast small files and learning resource. The resource distribution model is deployedin simulated situation through virtual machines on seven computers in the condition oflaboratory to test available data. The research designs the distribution model aiming at the features of learning resource,which can fit into the requirements of reliability, high efficiency, expandability and low costof resource storage. Through the simulation test in laboratory, the availability of this modeland the validity of storage strategy for small learning resource files can be verified.