Performance Study of Network Cache for Named Data Networking

Abstract

Caching strategy is widely adopted by current Internet to accelerate the process of content distribution. Compared to traditional caching technologies implemented in application layer of IP/TCP architecture, in-networking caching could provide better communication performance. Named Data Networking (NDN), a novel network architecture, has received significant attention recently. Benefiting from NDN named-based forwarding mechanism, in-network caching is naturally embedded in NDN. Previous studies have shown that NDN has enormous potential for end-user in latency reduction, throughput improvement, etc. However, there is a lack of knowledge about its performance gain in perspective of content provider. In this project, we designed and implemented a software simulator to investigate the benefit brought by NDN, in view of content provider. Series of simulations were conducted using NDN testbed network topology under various situations. Performance gain is measured in terms of the reduction of the workload of the content server. Better performance gain is obtained when (i) the cache hit probability at local CS is higher, (ii) the percentage of nodes with CS is increased and (iii) the percentage of CS-flow is raised. When above parameters are 20%, 80% and 60% respectively, the workload of the content server is reduced by around 25%.