CRII: NeTS: Configurable Optical Wireless Data Center Networks.

点击翻译按钮获取中文摘要

There is an increasing demand for efficient Data Center Networks (DCNs) to support the ever-expanding portfolio of data-centric applications that have a major role in enabling businesses, national security, Internet-of-Things (IoT), and social interaction. With today’s massive DCNs critical to companies such as Google and Amazon, to businesses such as the financial, communications and manufacturing industries as well as to government organizations, careful design of the DCN is critical to achieve efficient use of the server resources. Current DCNs use fixed cables to connect servers, leading to high cabling complexity and an inflexible topology that makes it difficult to efficiently provide the computation power across many applications. This project will explore the integration of Optical Wireless Communication (OWC) technology in DCNs to utilize the flexibility of wireless technologies to establish configurable on-demand links, and thus solve load hotspot and cabling complexity problems simultaneously. In addition to advancing the research in this area, this project will provide the opportunity to establish a network research track in the Department of Computer Science at Appalachian State University and will serve the needs of the diverse body of undergraduate and graduate students.This project will explore the design space of efficient, configurable OWC DCN topologies. Deploying wireless links in DCNs is challenging since high data-rate wireless links usually require, preferably, Line-of-Sight (LOS), point-to-point communication. The row-based physical topology used in conventional DCNs makes establishing point-to-point wireless links one of the most significant challenges facing wireless DCN designers. The research will focus on using graph theory and simulation to understand how different design parameters influence the performance of the DCN, and what mix of OWC and cable connectivity may provide the best performance. The preliminary results obtained from this project will serve as the guiding compass to realize a practical and efficient OWC DCN. The DCN simulation tool that will be developed by the PI and students, along with the results from this project will be made available to the research community.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.