Large-scale task processing for big data based on cloud computing has become a research hotspot nowadays. Many traditional task processing approaches in single domain based on cloud computing have been presented successively. Unfortunately, it is limited to some extent due to the type, price, and storage location of substrate resource. Based on this argument, a large-scale task processing approach for big data in multi-domain has been proposed in this work. While the serious problem of overheads in computation and data transmission still exists in task processing across multi-domain, to overcome this problem, a virtual network mapping algorithm based on multi-objective particle swarm optimization in multi-domain is proposed. Based on Pareto dominance theory, a fast non-dominated selection method for the optimal virtual network mapping scheme set is presented and crowding degree comparison method is employed for the final optimal mapping scheme, which contributes to the load balancing and minimization of bandwidth resource cost in data transmission. Cauchy mutation is introduced to accelerate convergence of the algorithm. Eventually, the large-scale tasks are processed efficiently. Experimental results show that the proposed approach can effectively reduce the additional consumption of computing and bandwidth resources, and greatly decrease the task processing time.
Virtual Network Mapping in Cloud Computing: A Graph Pattern Matching Approach
