Design And Implementation Of Distributed Storage System Based On Spatio-temporal Characteristic Data

Spatio-temporal data have the characteristics of space and time,which record the location and changed information with time of things in detail.In the era of information technology,Production business become more and more complex,and the spatio-temporal data have also become more widely used.From the image data captured by monitor in the smart city,to information about weather condition of different places over time,to log data generated by applications,these data are spatiotemporal data.With the widely used of spatio-temporal data,How to store and query these data efficiently has become a subject deserved to study.Compared with the traditional data types,spatio-temporal data have the characteristics of space and time.spatio-temporal data are usually generatred by the growth of time series.In the later stage,the upper application may use data mining and intelligent analysis technology to process data,so as to obtain the useful information.Spatio-temporal have the following characteristics: The data have a spatial attribute which indicates the producer of the data;Data generation is time-series,and the data are often continuously written to the storage systems;Upper application queries for spatiotemporal data are usually in large quantities;These is no modification to the data,which indicates the authenticity of things in a certain time and space.so how to implement an efficient storage system for spatio-temporal data is a subject that needs to be studied in this thesis.Base Based on a self-designed and developed distributed storage system for spatiotemporal data,this thesis expounds in detail the involved theoretical technologies,the design ideas,the implementation scheme,the testing and analysis of the system.The main work of the thesis is as follows.(1)The thesis deeply researchs the related theories and technologies of the distributed storage system;designs a node layout architecture of the distributed storage system with the way of central node.Central Node provides the services in a cluster mode of one master and multiple subordinates,which solves the single point of failure problem of the central node.The system architecture supports dynamic change of nodes.(2)Through the detailed analysis of the characteristics of spatio-temporal data,the thesis proposes an efficient storage strategy for spatio-temporal data based on LSM-Tree,according to the spatio-temporal and temporal characteristics of spatio-temporal data.According to the efficiency of disk sequential access,the system persists data in the form of data block files,and establishing distributed index for the data blocks to improve the efficiency of the data retrieval.The system divides the spatio-temporal data based on the data range slicing algorithm,which improves the efficiency of data query and the disaster tolerance of the system.(3)The detailed functional testing and performance testing of the distributed storage system are carried out in the thesis.The test items include node changes,data storage and data query,etc,and the test results are analyzed in detail.