AUTOMOTIVE

In a workshop hosted by the UIC Electronic Visualization Laboratory (EVL), NTT showed its MaXimal Queuing (MXQ) algorithm to leading network academics, and demonstrated its effectiveness at preventing congestion in highly-utilized networks. EVL's TeraVision was used to send three high-resolution video streams to a super-high-definition tiled display, while an Access Grid distributed video conferencing system also ran over the MXQ-enabled network. Attendees, some of whom participated via the Access Grid, were shown the signal degradation inherent in times of network congestion using conventional IP/MPLS routers, then the elimination of this congestion with new flow-based Caspian Apeiro routers incorporating NTT's MXQ algorithm. In flow-based networking, packets are routed as whole flows, i.e. streams of related packets, rather than as individual packets as in current IP/MPLS networks. The unique level of data obtained in flow-based routing, such as flow length, rate, delay variation and other parameters, enable a number of new network benefits. "NTT's MXQ flow-based control mechanism arbitrates flows based on their sending rate," said Dr. Takashi Shimizu of NTT Laboratories. "When congestion occurs, MXQ helps routers intelligently discard (or delay where possible) packets from higher rate flows. This allows the protection of smaller-rate flows and improves utilization of the network. This sort of congestion control intelligence is only possible in a flow-based environment, where routers know which packets belong to which flow." "MXQ's arbitration of flows according to their sending rate was achieved under realistic conditions using a combination of TeraVision streams to a 30-megapixel GeoWall-2 tiled LCD display, and Access Grid video conferencing" said Professor Jason Leigh, of EVL. "Caspian's MXQ implementation worked very well, even under real-world traffic." Today's demonstrations were part of the ON*VECTOR Advanced Networking Workshop, hosted at UIC/EVL, which focused on new flow-based networking techniques to improve QoS and enable better service models for IP-based telecommunications. They were the result of ongoing collaboration among NTT Labs, UIC/EVL and University of Tokyo. Four months of testbed experiments, conducted with the cooperation and technical support of Caspian Networks and NTT Communications, allowed researchers to gather experimental data on flow-based networking using the high-performance computational, networking and visualization infrastructure developed at EVL. Additional demonstrations by the University of Tokyo and Caspian Networks further illustrated the benefits of flow-based networking, including peer-to-peer traffic identification and control, as well as denial-of-service (DoS) prevention. Workshop participants included attendees from UIC, NTT Laboratories, NTT America, NTT Communications, the University of Tokyo, Caspian Networks, Northwestern University, and Argonne National Laboratory.