The Research On Data Replica Management Strategy In Cloud Computing

The development of Internet, in a large measure, stimulates the explosive growth of global data, while lousy ability of data processing, to a certain extent, restricts the progress of the whole industry. To get rid of the bad reliability and low security of data, uneven load of nodes and so on, data replica management strategy comes to be some hot spot.A good replica management strategy can shorten response time, decrease occupied bandwidth, and balance the node load, further, improve the QoS of clouding and users’ satisfaction.Based on traditional data replica management strategies, this paper summarizes their limitations and along with the challenges network nowadays faces, it mainly discussed the data replica selection strategy and the data replica consistency strategy. The main work of this thesis includes:1.Analyze the research status and basis of replica selection strategies, and put forward the replica selection strategy based on ant colony algorithm, which includes in detail the setting of pheromone initial value, correction of replica selection probability and improvement of pheromone update formula.Through the simulation on it, in comparison with the traditional replica selection strategies, ant colony algorithm can not only make timely reaction to end-users, but also balance the load of nodes to enhance the overall performance of the system.2.Analyze the research status of replica consistency strategies, elaborate the maintenance characteristics and criterions of replica consistency, and propose the strategy of RCS-UAL based on update rate, access frequency and node load, which includes in detail real-time updates of main replicas under the three-layer frame, update by category for secondary replicas, main replicas pushing the update information to secondary replicas.Through the simulation on it, it can be shown that on the general performance of response time to end-users and copy number, RCS-UAL can make a good performance.