Cache Performance Investigation in NDN for Large Scale Scientific Data

Abstract

Named-Data Networking (NDN) is the most active instance of Information-centric Networking (ICN) having a very active research community. NDN is racing for the future Internet architecture by removing dependence on IP addresses, location, and host-to-host communication model. The communication paradigm in NDN revolves around the content/data by naming the content, securing it instead of the channel, and retrieving the content by its names and not the IP address of the machine that hosts the data. NDN has lots of benefits for large scale scientific data as it has very expressive naming support, access control, and enhanced delivery performance compare to traditional IP based networking. NDN supports in-network caching that highly enhances the performance of NDN by reducing latency and network congestion through retrieving popular contents from nearby caches. Considering its importance, we in this work investigate the performance of in-network caching for large scale scientific data. We infer from our simulation results that the optimal cache size and the percentage of the cache hit depend on multiple parameters i.e. traffic pattern, traffic load, and cache replacement policies as proved with simulation results.