一种基于代理的协同内容分发系统的体系结构设计
An Architectural Design for Brokered Collaborative Content Delivery System
摘要：
网络技术的进步推动了大量内容的上传和请求，这些内容可以通过多媒体网络和社交网络服务的增加使用来识别。这种情况迫使内容提供方扩展其基础设施，以便应对网络流量、存储和其他资源的额外供应。由于在扩展基础设施的复杂性和成本因素的存在，我们提出了一种新的解决方案，通过引入一个代理的协作内容交付系统，为客户端提供和交付内容。该系统的体系结构设计利用了其他内容提供方中的内容冗余和内容分发机制来向客户端递送内容。随着最近云计算的出现，我们表明，该系统也可以采用在云上运行。在本文中，我们专注于一个中介计划，以调解用户的请求，最合适的内容提供方的基础上排名系统。该体系结构提供了一种新的全球排名值（GRV）的概念，估计内容提供方的能力和转换的QoS要求的内容请求。一个公平的模型，这将使这种设计是有吸引力的，目前的内容交付制度也被引入。通过模拟，我们表明，使用公平的供应商选择，内容可以提供一个更好的池合格的供应商，从而利用合作，防止潜在的QoS违规时，可能会发生池的大小较小。
英文摘要：
Advances in web technologies have driven massive content uploads and requests that can be identified by the increased usage of multimedia web and social web services. This situation enforces the content providers to scale their infrastructure in order to cope with the extra provisioning of network traffic, storage and other resources. Since the complexity and cost factors in scaling the infrastructure exist, we propose a novel solution for providing and delivering contents to clients by introducing a brokered collaborative content delivery system. The architectural design of this system leverages content redundancy and content distribution mechanisms in other content providers to deliver contents to the clients. With the recent emergence of cloud computing, we show that this system can also be adopted to run on the cloud. In this paper, we focus on a brokering scheme to mediate user requests to the most appropriate content provider based on a ranking system. The architecture provides a novel Global Rank Value (GRV) concept in estimating content provider capability and transforming the QoS requirement of a content request. A fairness model that will bring this design to be attractive to the current content delivery regime is also introduced. Through simulation, we show that using fair provider selection, contents can be provisioned by a better pool of qualified providers thus leveraging the collaboration and preventing potential QoS violation that may occur when the size of pool is smaller.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
