Study On The Network Rules Compression And Finite Element Parallel Computing Algorithm Based On Cloud Computing

Cloud computing is a new computing model after distributed computing,grid computing and peer-to-peer computing.The core idea of cloud computing is resource renting,application hosting and outsourcing services.Providing users with a flexible,secure and efficient network application platform is a key issue in cloud computing.In this paper,we studied the compression of routing rules in cloud computing and finite element parallel computing in cloud platform respectively.Include the following aspects:1.In OpenFlow network,supporting real-time rule updating in limited storage capacity of TCAM is a key challenge for OpenFlow.In this thesis,a new approach is presented to solve this problem.In our approach,the TCAM chip is divided into two areas,a real-time update area and a compression storage area.The former is assigned in the front of the chip for storing real-time updating rules sent by the controller;the latter is used for storing compressed rules generated by the server within certain time period.We have made a comprehensive analysis on the space division ratio and conducted simulation experiments on the rules generated by the ClassBench tool.The experimental results demonstrated the efficiency and effectiveness of our approach.2.We propose a parallel computing algorithm based on finite elements.The basic idea of our algorithm is to break down a large task into multiple sub-tasks based on the method of finite elements.And each of these sub-tasks is designed for computing a partition of the original large task.The original large task's result will be computed from all these sub-tasks' results.First,we decide the value to the unit number for the finite element model and then according to the value of unit number to divide finite elements.In this model,each unit corresponds to a finite element subtask.After that,we schedule all sub-tasks and assign each sub-task to the node with the lowest weight,which is composed by the expected response time and computing cost.We conduct the EBE method to complete the parallel computing.The last step is to compute for each unit the stiffness matrix,load vector and boundary conditions.Then the displacement and stress of each unit are computed in parallel,the overall nodal displacement and stress computation are based on the results of each sub-unit.In order to verify the effectiveness of the proposed algorithm,we conducted the simulation experiment based on real heterogeneous system,and the experimental result shows that the proposed algorithm can achieve great speedup and high parallel efficiency.