Edinburgh Napier University

  High quality online video streaming, both live and on-demand, has become an essential part of consumers' every-day lives. The popularity of video streaming has placed a heavy burden on the network infrastructure that now has to transfer an enormous amount of data very quickly to the end-user. To further exacerbate the situation, the Video-on-Demand (VoD) distribution paradigm uses a unicast independent flow for each user request. This results in multiple duplicate flows carrying the same video assets many times end-to-end. We present OpenCache: a highly configurable, efficient and transparent in-network caching service that aims to improve the VoD distribution efficiency by caching video assets as close to the end-user as possible. OpenCache leverages Software Defined Networking to benefit last mile environments by improving network utilisation and increasing the Quality of Experience for the end-user. Our evaluation on a pan-European OpenFlow testbed uses adaptive video streaming and demonstrates that with the use of OpenCache, the external link utilisation is reduced by 100%. Furthermore the streaming application receives better quality video and observes higher throughput, lower latency and shorter start up and buffering times.