The Design And Implementation Of Multilevel Storage System Oriented To Multidimensional Data

With the rapid development of Internet of Things(IoT),the application of IoT is also becoming more and more widespread.Many devices gain access of IoT,each of which will provide specialised type of data.While with the growing number of data to be stored,single storage can no longer meet the data storage requirements.Moreover,current data storage mode makes data flat,disconnecting the relation of different data dimensions.In addition,the single storage mode makes it difficult to make full use of data to carry out further data analysis and data mining.In order to solve these problems,this paper designs and implements a multi-level data storage system for IoT applications.The first level of storage applies relational database to store the data of sensor device after parsing,thus providing support for data queries of situation monitoring system.And the first-level storage system will synchronize the data to the second-level distributed data storage,which works as data warehouse with full set of data.Besides,the second level of storage will support large-scale data processing and make it realizable that distributed storage has the ability of disaster recovery and backup mechanism.After data processing using the model of MapReduce according to different dimensions,the third level will store the data by dimensions.Firstly,this paper analyzes the existing problems in the storage system,which is using now.Secondly,it proposes a solution to each problem in the system and determines the structure of multi-level storage system and the ways to achieve.Then,this paper implements the multi-level storage system according to the scheme.Finally,it tests the function of system and performance indicators.