论文部分内容阅读
越来越多的新型万维网(Web)应用开始提供在线音视频交流功能,而有效的Flash媒体服务器部署方案是提高服务质量的基础。该文给出了媒体服务器优化部署问题的数学模型,并证明即使用户需求和网络性能信息已知,求解最优部署方案仍然是NP-hard问题。提出了优化部署方案的贪婪算法和分别基于重要程度和拓扑特征的启发式算法,分析了它们的优缺点和复杂度。基于互联网的真实拓扑的实验表明:该几种算法的效果都明显优于随机部署,而贪婪算法和基于重要程度的启发式算法甚至能在有效节省成本的前提下,接近全局部署的效果。
More and more new Web applications begin to provide online audio and video communication capabilities, and an effective Flash media server deployment solution is the basis for improving service quality. This paper presents a mathematic model of media server optimization deployment problem and proves that solving optimal deployment scheme is still an NP-hard problem even though user requirements and network performance information are known. The greedy algorithm to optimize the deployment scheme and the heuristic algorithm based on importance and topological features are proposed respectively, and their advantages, disadvantages and complexities are analyzed. Experiments based on the real topology of the Internet show that these algorithms are significantly better than random deployments, whereas greedy algorithms and heuristic algorithms based on importance can approach the effect of global deployment even with cost savings.