An Internet-Wide Distributed System for Data-Stream Processing

The ubiquity of the Internet has stimulated the development of data- rather than processor-intensive applications. Such data-intensive applications include streaming media, interactive distance learning, and live web-casts. While plenty of research has focused on the real-time delivery of media streams to various subscribers, no solutions have been proposed that provide per-subscriber QoS guarantees to hundreds of thousands of subscribers distributed on an Internet scale. Moreover, there has been no successful approach that integrates heterogeneous end-hosts into a distributed system, connected via an overlay network, for the purpose of QoS-constrained data processing and delivery. This paper proposes a pipelined distributed system that is the foundation for the delivery of QoS constrained datastreams over a scalable backbone consisting of heterogeneous commodity-off-the-shelf (COTS) systems. To provide a scalable routing substrate that can adapt to QoS demands and constraints, a peer...

• Type: Conference
• Year: 2004
• Where: IC
• Authors: Gabriel Parmer, Richard West, Xin Qi, Gerald Fry, Yuting Zhang
• Keywords: Distributed System, IC 2004, IC 2007, Per-subscriber Qos Guarantees, QoS Constrained Datastreams