Discuss (i) how the research adds to the understanding of the area investigated; (ii) the technical effectiveness and economic feasibility of the methods investigated; and (iii) how the project is otherwise of benefit to the public. The discussion should be written in terms understandable by an educated layman Research-and-Education Network (REN) providers have deployed both connectionless IP-routed and dynamic circuit networks. The objective of this project is to determine how these two networks can be integrated in a manner that offers both resource optimization and user-performance improvements. This will allow an integrated packet/circuit network to scale better to multi-Terabits/sec capacities as will soon be required to support the high-performance networking needs of DOE scientists. The first solution investigated, Hybrid Network Traffic Engineering System (HNTES 1.0), was an online method for identifying long-duration flows arriving on the IP-routed network, and then setting up a dynamic circuit to redirect these flows to the high-speed circuit network. We changed our focus to size from duration after recognizing that long-duration flows may not be high in the rate dimension. Instead, high-rate flows are often generated by data transfer nodes dedicated to moving large-sized datasets, and high-rate flows are the ones that can have adverse effects on delay-sensitive flows. Furthermore, the need to port mirror packets to a server cluster in HNTES 1.0 for online detection of heavy-hitter flows could be cost-prohibitive.
PDF
download
878987797
028-85220240
