Experimental evaluation of Linux TCP for adaptive video streaming over the cloud

With the popularity of video streaming on the Internet an entirely new type of media player has been created called the adaptive video player that dynamically adjusts video quality depending on available network bandwidth. When combined with cloud computing this new way of serving media has the potential to change the video streaming landscape by allowing content to be streamed reliably from anywhere in the world. In this paper we evaluate the use of different TCP congestion control algorithms over the Cloud including five major variants: Cubic, Bic, Vegas, H-TCP, and HighSpeed TCP. Both private (based on Eucalyptus) and public (Amazon EC2) cloud environments are considered. We evaluate video streaming performance, TCP friendliness, and for each criterion identify the top performer. We believe that this is an important step towards optimizing TCP or another transport protocol for supporting cloud computing.