Threshold-Based Dynamic Replication in Large-ScaleVideo-on-Demand Systems

  • Authors:

  • Peter W. K. Lie;John C. S. Lui;Leana Golubchik

  • Affiliations:

  • Department of Computer Science & Engineering, The Chinese University of Hong Kong, Hong Kong;Department of Computer Science & Engineering, The Chinese University of Hong Kong, Hong Kong;Department of Computer Science and UMIACS, University of Maryland, College Park, MD, 20742

  • Venue:
  • Multimedia Tools and Applications
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

Recent advances in high speed networking technologies andvideo compression techniques have made Video-on-Demand (VOD) servicesfeasible. A large-scale VOD system imposes a large demand on I/Obandwidth and storage resources, and therefore, parallel disks aretypically used for providing VOD service. Although striping of moviedata across a large number of disks can balance the utilization amongthese disks, such a striping technique can exhibit additionalcomplexity, for instance, in data management, such as synchronizationamong disks during data delivery, as well as in supporting faulttolerant behavior. Therefore, it is more practical to limit theextent of data striping, for example, by arranging the disks ingroups (or nodes) and then allowing intra-group (or intra-node) datastriping only. With multiple striping groups, however, we may needto assign a movie to multiple nodes so as to satisfy the total demandof requests for that movie. Such an approach gives rise to severaldesign issues, including: (1) what is the right number of copies ofeach movie we need so as to satisfy the demand and at the same timenot waste storage capacity, (2) how to assign these movies todifferent nodes in the system, and (3) what are efficient approachesto altering the number of copies of each movie (and their placement)when the need for that arises. In this paper, we study an approachto dynamically reconfiguring the VOD system so as to alter the numberof copies of each movie maintained on the server as the access demandfor these movies fluctuates. We propose various approaches toaddressing the above stated issues, which result in a VOD design thatis adaptive to the changes in data access patterns. Performanceevaluation is carried out to quantify the costs and the performancegains of these techniques.