Research On Data Centers Storage Resource Allocation Strategy For Data Of Users In Social Network

With the rapid development of the Internet,online social networks have gradually influenced people's lives and become an indispensable means of communication for modern people.People can use text,pictures,videos,etc.to share their opinions and feelings with friends all over the world through online social networks at any time and place.At the same time,users often browse the status shared by friends and obtain their data.Different with traditional web applications,data in online social networks are almost entirely generated by users.With the addition of a large number of users,the data in online social networks have shown explosive growth.However,due to the limited storage space,a single data center cannot store all data.It is necessary to divide data into different data centers.How to allocate appropriate storage resources for data to reduce the response delay of user data acquisition and ensure online social networks operation efficiency is of utmost importance.Storage resources are allocated based on the community structure of social networks in this thesis.A clustering algorithm is applied to divide the community and user's geographical location and data centers load status are combined to manage data centers storage resources.The main research work of the thesis includes:(1)Due to the low applicability and performance of the traditional community partitioning algorithm,the K-medoids clustering algorithm is applied to the community division.Firstly,the similarity and intimacy between users are calculated based on the topological characteristics of social networks and the user's behavioral information in time and space,and the measurement mechanism of multi-dimensional social distance between users is given.Then the multi-dimensional social distance is applied to the K-medoids clustering algorithm.(2)The data of users in online social networks are divided into different data centers according to the result of the community division,then copies of users and friends will be created when they are distributed in different data centers.A Community Based Data Placement Algorithm(CBDPA)is proposed to ensure data fault tolerance and reduce data acquisition response delays and copy numbers.Besides,in order to adapt to dynamic changes of social networks,a Dynamic Adjustment Data Placement Algorithm(DADPA)is proposed to dynamically adjust the placement of user's data based on changes in user status and friend relationships.(3)In order to verify the performance of the proposed algorithm,the storage resource allocation and dynamic adjustment experiments are carried out and compared on the real Foursquare dataset.The results show that:(a)With the same number of copies,the average response time and data centers load of the CBDPA algorithm are reduced by more than 50% compared with the other algorithms compared in this thesis.(b)The dynamic adjustment algorithm proposed in this thesis can reduce the increase in response latency caused by changes in social networks.