NDM 2012‎ > ‎

Invited Talk


Title:  "Optimizing Transport of Big Data over Dedicated Networks"  [slides]

Abstract
Data centers are being deployed in cloud computing environments, scientific, financial, defense, and other enterprises.Geographically distributed data centers transmit and receive growing volumes of data. In order to avoid congestion in the public internet, they use high speed dedicated optical networks, which can be thought of as private highways for carrying data. A careful examination of the impact of such high speed network traffic on a commodity multicore machine show packet loss and degraded throughput due to the end-system being the bottleneck.We found that high speed single flow traffic nullifies the benefits of multicore systems and multiqueue NICs. We propose an end-system aware flow control technique to optimize the data transfer time using rate based protocols. Using introspective end-system modeling, we show that we can determine the optimal number of parallel flows required to utilize the available bandwidth and the optimal rate for each of the flows. 




Speaker:    Dipak Ghosal

                    Professor
                    Department of Computer Science
                    University of California, Davis




Prof. Dipak Ghosal received the B.Tech degree in electrical engineering from the Indian Institute of Technology, Kanpur, India, in 1983, the M.S. degree in computer science from the Indian Institute of Science, Bangalore, India, in 1985, and the Ph.D. degree in computer science from the University of  Louisiana, Lafayette, in 1988. From 1988 to 1990, he was a Research Associate at the Institute for Advanced Computer Studies, University of Maryland at College Park. From 1990 to 1996, he was a Member of Technical Staff at Bell Communications Research/Bellcore, (presently Telcordia), Red Bank, NJ. Since 1996, he has been with the faculty of  the Department of Computer Science, University of California at Davis, His research interests include control and management of high-speed networks, wireless networks, and performance evaluation of computer and communication systems.