Research And Implementation Of Industrial Time Series Database System Based On HBase

With the acceleration of informatization, intelligent devices have integrated into theworld in every aspects and the Internet of Things is guadually affecting our everyday life.In the past, the Internet of Things was mainly concerned about the objects which use radiofrequency identification technology to communicate with the Internet, and now, lots ofsensors and intelligent terminals are widely applied to industrial area. Massive amounts ofdata which generated by industries and its special data characteristic brings unprecedentedchallenge to the existing database systems. How to effectively manage industrial databecomes a serious problem. In order to meet the growing number of data capacity, we canrefer to the world’s widely used distributed database technology and industrial dataprocessing technology to study this issue.Traditional distributed database which based on data partitioning techniques can notguarantee the efficiency of storing and querying data, in the mean time, frequentcommunication between nodes has been a great consumption of the cluster, thus thetraditional distributed database can not bring an effective data management system forindustrial data. After a lot of research, this article mainly discusses the research and designof the industrial database based on the open source framework HBase. The high reliability,high fault tolerance and high scalable of HBase greatly reduce the difficulty of systemdevelopment.Industrial data is not only a large number of data but also has characteristics of timeseries data. The system is designed primarily considering the three aspects of real-timedata monitoring, data persistence and consistency of metadata. The whole industrial timeseries database system which uses real-time historical database as adaptation layer is buildon HBase.This article analyzes the key problem to design a system by organizing data storagewith efficient row/column key, using smooth data persistence strategy to alleviate clusterload and using ZooKeeper to guarantee the consistency of all data. At last, this article teststhe performance of this system according to the needs of application environment. The testresult shows that the data storage efficiency is very high, and it also illustrate that thesystem can efficiently handle the writing pressure and queries have excellent performance.