Mohammad Alian

University of Kansas

Intelligent Data Placement in the Terabit Per Second Network Era

Abstract

Terabit-per-second network connectivity is on the horizon. To handle such massive network traffic, intelligent data placement on the CPU chip is required. In this presentation, we provide insights into the microarchitectural requirements for handling network traffic in the terabit-per-second era. We then discuss an intelligent data placement technique for network data called IDIO. IDIO implements three synergistic technologies to steer received network data to different levels of the memory hierarchy and remove unnecessary data movement while processing packets on the CPU. IDIO significantly reduces data movement and consequently improves performance isolation and energy efficiency of future network-intensive servers.

Biography

Dr. Alian is an Assistant Professor in the Electrical Engineering and Computer Science Department at the University of Kansas since August 2020. His research group is developing novel hardware and software technologies to minimize data movement in next-generation large-scale computer systems. Dr. Alian obtained his Ph.D. from the Department of Electrical and Computer Engineering at the University of Illinois Urbana Champaign in July 2020. His research has been recognized by an NSF CAREER Award, a Miller Scholar Award, four best paper candidacies in top computer architecture conferences, and an honorable mention in IEEE Micro Top Picks. More information is available on his webpage, http://alian-eecs.ku.edu