The large size, hard real-time constraints, and general complexity of a video-on-demand system lead to many difficult design decisions that may affect the system’s ultimate cost and performance substantially. The primary design problems of the video-on-demand system are real-time disk scheduling, buffer management, perfecting, movie allocation, disk striping, device bandwidth reduction and others, where the movie allocation determines the number of movie copies and the placement of movie copies in video servers.

The movie allocation among these problems is important because it controls an efficient usage of space and bandwidth on video servers. In this paper, we study the movie allocation for distributed video-on-demand systems, and propose a movie replication algorithm considering space and movie popularity, and a movie allocation algorithm which has a new movie selection policy, grouped alternation and a new movie placement policy, SCAN.

We design two movie allocation algorithms: grouped alternate round-robin and grouped alternate SCAN, where the proposed algorithms allocate a group of movies instead of one movie to video servers at each round.

The performance of the proposed algorithms is evaluated through simulations. Simulation results show that the performance of the proposed algorithms is better than that of other algorithms in terms of load balancing among video severs’ bandwidths and blocking probability for a movie request.