Research On Optimization Of Dynamic Data Storage Based On Data Temperature

Cloud computing is an indispensable Internet infrastructure of the new generation,which greatly supports and accelerates the development of big data and artificial intelligence.As the extension of cloud computing technology,edge computing provides computing and storage services with high bandwidth,low latency and high security,which has become a beneficial supplement to cloud computing.Under the background of the interconnection of things and the generation of massive data,the demand for data computing and storage is growing rapidly,relying more and more on the resources provided by cloud computing and edge computing.In the face of complex cloud market,the charging items,pricing strategies and performance of cloud servers belonging to various providers are very different.In addition,the single cloud has many risks such as vendor lock-in,data leakage,and low availability,which makes multi-cloud storage be the development trend.However,how to choose the appropriate cloud services to meet the needs of users and make some Qo S indicators achieve the optimum are challenges and problems to be solved.Furthermore,edge environment has great heterogeneity and volatility,which brings about more restrictive,makes it more complex in application scenarios,and is more difficult to deal with above problems.Most data in the real world have spatial and temporal attributes,and some other essential data attributes also have temporal and spatial variability.In the research field of cloud computing and edge computing,these features greatly impact data placement.However,these critical spatiotemporal features have been largely ignored by existing studies.Therefore,aiming at the above challenges,this paper focuses on the optimization of data storage based on its spatiotemporal attributes in edge-cloud environments.The main work of this paper includes the following parts:(1)Synthesizing the time-space-related attributes of data,abstracting the concept and definition of "data temperature",and giving corresponding models and calculation methods.The method takes the change of time as the main reference basis,calculates the temperature value and obtains its distribution by integrating the geographical distribution,density of requests,the average number of users' visits,the popularity range,and the degree of concentration,then reckons the regional temperature and global temperature.Data temperature is the basis of the research in this article,supporting through the subsequent research on data optimization storage strategies.(2)Aiming at the problem of dynamic data optimization storage in multi-cloud environments,this article firstly gives the relevant definition of the problem,and then based on the proposed data temperature model,obtains the corresponding cloud service set by temperature distribution in different regions.In this paper,weight method and storage scheme score are used to convert the multi-objective optimization into the single-objective optimization problem,and applies the ant colony algorithm to acquire the optimized data storage scheme.On this basis,this paper discusses the strategy of dynamic adjustment when the data temperature and cloud service changes in order to achieve rapid response and optimization of Qo S indicators throughout the data storage cycle.(3)In view of the data optimization storage problem in edge-cloud environments,this paper mainly considers the cost optimization under the delay constraint.Firstly,this chapter gives the definition of related problems and optimization goals.Then according to the temperature distribution,various numbers of copies are placed in different areas.This paper uses the 0-1knapsack model to solve the problem,and takes advantage of dynamic programming algorithm to obtain the best placement solution.Experiments show that the temperature-based method not only can save more cost compared with the classical data placement strategies,but also exerts the important value of the data by storing data in its popular areas.