TY - GEN
T1 - Support for multiple classes of traffic in multicomputer routers
AU - Rexford, Jennifer
AU - Shin, Kang G.
N1 - Funding Information:
* The work reported in this paper was supported in part by the National Science Foundation under Grant MIP-9203895 and an Office of Naval Research graduate fellowship. Any opinions, findings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of the funding agencies.
PY - 1994
Y1 - 1994
N2 - Emerging parallel real-time and multimedia applications broaden the range of performance requirements imposed on the interconnection network. This communication typically consists of a mixture of different traffic classes, where guaranteed packets require bounds on latency or throughput while good average performance suffices for the best-effort traffic. This paper investigates how multicomputer routers can capitalize on low-latency routing and switching techniques for best-effort traffic while still supporting guaranteed communication. Through simulation experiments, we show that certain architectural features are best-suited to particular performance requirements. Based on these results, the paper proposes and evaluates a router architecture that tailors low-level routing, switching, and flow-control policies to the unique needs of best-effort and guaranteed traffic. Careful selection of these policies, coupled with fine-grain arbitration between the classes, allows the guaranteed and best-effort packets to share network bandwidth without sacrificing the performance of either class.
AB - Emerging parallel real-time and multimedia applications broaden the range of performance requirements imposed on the interconnection network. This communication typically consists of a mixture of different traffic classes, where guaranteed packets require bounds on latency or throughput while good average performance suffices for the best-effort traffic. This paper investigates how multicomputer routers can capitalize on low-latency routing and switching techniques for best-effort traffic while still supporting guaranteed communication. Through simulation experiments, we show that certain architectural features are best-suited to particular performance requirements. Based on these results, the paper proposes and evaluates a router architecture that tailors low-level routing, switching, and flow-control policies to the unique needs of best-effort and guaranteed traffic. Careful selection of these policies, coupled with fine-grain arbitration between the classes, allows the guaranteed and best-effort packets to share network bandwidth without sacrificing the performance of either class.
UR - http://www.scopus.com/inward/record.url?scp=84893777657&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84893777657&partnerID=8YFLogxK
U2 - 10.1007/3-540-58429-3_32
DO - 10.1007/3-540-58429-3_32
M3 - Conference contribution
AN - SCOPUS:84893777657
SN - 9783540584292
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 116
EP - 130
BT - Parallel Computer Routing and Commumcation - 1st International Workshop, PCRCW 1994, Proceedings
A2 - Bolding, Kevin
A2 - Snyder, Lawrence
PB - Springer Verlag
T2 - 1st International Workshop on Parallel Computer Routing and Communication, PCRCW 1994
Y2 - 16 May 1994 through 18 May 1994
ER -