Pervasive Streaming via Peer-to-Peer Networks

Video Delivery Platforms

The Client –Server Model is the opposite extreme of Peer-to-Peer Computing. However, the latter can be viewed as the evolution of the former. In a CS-based streaming setting, the client initiates a connection with the video source address and the server replies back by directly delivering the content. Though simplicity and manageability are two major advantages of this scheme, its weaknesses pushed its evolution to P2P Networks. As all the content is located and provided by a single central entity, any failure of that entity may deactivate the whole streaming service to any client. There is a plethora of unpredictable either accidental (e.g. power cut-offs) or deliberate (e.g. security attacks) reasons for these failures. Traffic bursts are also conditions that these architectures are not designed to handle. Sudden increases of content requests from clients can quickly consume all the resources of the server and force it to drop any excessive load. In an effort to tackle those problems, system administrators increased the initial investment and maintenance costs by building very powerful, highly secured infrastructures that only very specialized personnel could operate. They quickly became unscalable systems that few could afford (Androutsellis-Theotokis & Spinellis 2004; Liu et al. 2008) and without even providing guarantees for service quality and reliability in case of flush crowds. The need for a more distributed architecture drove the development of Content Delivery Networks (CDNs) (Pallis & Vakali 2006).