互联网流媒体的多边微监控
Multilateral Micro-Monitoring for Internet Streaming
摘要：
视频流正在主导互联网。为了与传统有线电视和卫星电视的性能竞争，内容提供方将内容交付外包给第三方内容分发网络和经纪人。然而，没有现有的监控机制提供流媒体服务的性能的多边视图。换句话说，没有审计机制反映内容提供方、内容分发方和最终用户关于服务执行情况的相互协议。在本文中，我们提出了UgoVor，一个系统，用于监测多边流媒体合同，这是可执行的描述之间的相互协议的内容提供方，内容分销商和最终用户。我们的关键见解是，实时多边微监控-能够解释每个重新缓冲事件和流中每个视频块的分辨率-不仅是可行的，而且是一项互联网可扩展的任务。为了证明这一说法，我们在一个为期10个月的实验中评估了UgoVor，该实验对应于超过25年的流数据，包括来自1,300多个独特AS的客户端的超过430，000个流会话。我们的测量证实，UgoVor可以提供一个准确的分布式性能共识的互联网流，并可以帮助从根本上推进现有的性能不可知的定价模型，以新颖和透明的支付你的经验。
英文摘要：
Video streaming is dominating the Internet. To compete with the performance of traditional cable and satellite options, content providers outsource the content delivery to third-party content distribution networks and brokers. However, no existing monitoring mechanism offers a multilateral view of a streaming service's performance. In other words, no auditing mechanism reflects the mutual agreement of content providers, content distributors and end-users alike about how well, or not, a service performs. In this paper, we present UgoVor, a system for monitoring multilateral streaming contracts, that is enforceable descriptions of mutual agreements among content providers, content distributors and end-users. Our key insight is that real-time multilateral micro-monitoring -- capable of accounting for every re-buffering event and the resolution of every video chunk in a stream -- is not only feasible, but an Internet-scalable task. To demonstrate this claim we evaluate UgoVor in the context of a 10-month long experiment, corresponding to over 25 years of streaming data, including over 430,000 streaming sessions with clients from over 1,300 unique ASes. Our measurements confirm that UgoVor can provide an accurate distributed performance consensus for Internet streaming, and can help radically advance existing performance-agnostic pricing model towards novel and transparent pay-what-you-experience ones.
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。
