EFFICIENT DATA STORAGE AND RETRIEVAL TECHNIQUES FOR HIGH-PERFORMANCE VIDEO SERVERS
Jinsung Cho · 2000
During the past decade we have paid great attention to a video-on-demand service which provides the combined facilities of a video rental store over high-speed networks. The realization of such services requires the development of video servers that support efficient mechanisms for storing and retrieving video data. The study on video servers is highly motivated by the fact that video-on-demand services are becoming increasingly important and widespread in the entertainment, education, and telecommunications industries. In this thesis we design and/or implement video servers in various environments by providing efficient storage and retrieval of video data. This thesis consists of three main parts: storage and retrieval in a single server, in a large-scale server, and in a multi-resolution video server. In the first part, we begin by designing and implementing a small-scale single server. The large size and bandwidth of video data require a single server approach to adopt disk arrays for its storage subsystem. Through a simple performance analysis for disk arrays, a prototype of the small-scale video server is developed. The second part of this thesis explores the issue on designing a large-scale video server. Limitations in scalability of the single server approach lead to a