A Transparent Distributed Computing Framework And Its Network Optimization Software Applications

It refers to substantial calculation in the study of network optimization and network simulation, such as routing prediction, geodesic calculation, and so on. And the calculation usually spends a lot of memory and CPU resources. It will take a lot of time if you use a single computer to deal with this enormous amount of data and calculation, usually at least several hours, which is a waste of time for the network planner and network simulation persons. But it is too expensive of supercomputers, and it have a lot of difficulties to configure cluster system, such as high cost, technical complexity, hardware requirements and the limited number of nodes. Under this circumstances, this article made the study subject, it make full use of he existing network equipment and idle computer resources, distribute the volume of stand-alone computing to many machines for parallel processing, which greatly shorten the computing time, and to an utmost degree, shield the bottom of distributed computing process so as to make it easy to use and configure.In order to deal with the problem of a large quantity of data, a large volume of computing and slow speed of stand-alone computer in the data-processing of network optimization and network simulation, this article proposed a distributed processing scheme, which can make full use of existing resources, get a good effect and strong ability of network reconfiguration, and can easy upgrades, easy to use. The system provides a unified interface which can adapt to all Killds of parallel algorithm by using DLLs. The system has a higher intelligence, is able to automatically detect the machine information and network environment and it make more convenience for the user to configure the system. Through the reasonable task scheduling, the distributed mission of high-performance computing nodes are more than low-performance computing nodes. It uses hierarchical system architecture, supports the inter-department networKillg. This system uses real-time management of the status of the resources use of network nodes to realize the load balance, makes full use of network resources, and reduces response time of business processes. In the view of application, our research of this dissertation enable small departments with dozens of computers, such as laboratories, to become powerful distributed network computing platforms , while it only requires the computers have access of networks without additional hardware. These platform can perform quick of sophisticated system computing problems, furthermore, they does affect computers' daily use. As for bigger computing taks, they can be divided into sub tasks which will be computed in networked PCs. With ever-enlarging networks, it can improve development efficiency, vastly transforming computing patterns in institutions and enterprise. Moreover, it has potential enconomic benefits.This dissertation proposes LAN-based distributed computing for network performance tuning. In limited economic situation, we adopt distributed computing to solve those problem.