Jolley Hall, Room 309
"Performance Measurement and the Guided Improvement of Software Routers"
Adviser: Patrick Crowley
This dissertation proposal aims to improve software routers in Information Centric Networking. Information Centric Networking (ICN) is an exciting new Internet architecture that is built on a simple concept: the request for named data. The network makes hop-by-hop decisions based on the name of the data, as opposed to an IP address bound to a specific location. This architecture better matches the way we use the Internet today—we do not care where our content came from, so long as we can verify it is what we asked for. While ICN has seen a fair amount of interest from academia, the performance characteristics of ICN software routers for consumers has not been widely pursued. In order for ICN to be successful, it needs to operate seamlessly and provide benefits at the consumer level. However, the performance landscape for ICN software routers is unknown. In fact, there is no standard methodology for testing the performance of ICN software routers. Due to key data structure differences, tools and techniques from IP cannot be re-used without modification. We aim to develop a thorough performance testing methodology for ICN software routers, using statistical techniques that proved effective when analyzing Linux software routers (a well studied system). This dissertation will culminate with a best-of-breed ICN software router, based on conclusions from the ICN software router performance analysis. The proposed ICN software router will focus on performance at consumer devices, e.g. a laptop or raspberry pi, as opposed to well-provisioned servers—success in this domain is crucial for the adoption of ICN.