A Model of Job Scheduling with Deadline for Video-on-Demand System

Scalability and high degree of playing continuity are key points to large-scale application of video-on-demand system. This paper presents a peer-to-peer (P2P) video-on-demand system CPVoD based on a cylindrical structure. It can enhance the efficiency of finding data by combining the accuracy and efficiency of DHT and the simplicity and practicability of Gossip protocol. A data scheduling strategy is introduced which starts from the perspective of reducing the expired data. By adding a small amount of local information to the data information exchange process based on Gossip protocol, it can help peers find the optimal download resource and finally enhance the playing continuity of system. A data downloading mathematical model is introduced, which converts the problem of improving the playing continuity of VoD system to the problem of finding the optimal solution for job scheduling with deadline, and a corresponding solution algorithm is also presented. The emulation experiments show that CPVoD can guarantee more than 99% of nodes joining the system normally and obtain a playing continuity more than 95% on the condition that server load is stable.