DVDs: Much Needed “Shot in the Arm” for Video Servers

  • Authors:
  • Vijnan Shastri;P. Venkat Rangan;Srihari Sampath-Kumar

  • Affiliations:
  • CEDT, Indian Institute of Science, Bangalore-560 012, India. E-mail: vshastri@cedt.iisc.ernet.in;Department of Computer Science and Engineering, University of California, San Diego, CA, 92093-0114, USA. E-mail: venkat@cs.ucsd.edu;Department of Computer Science, Washington University, St. Louis, MO 63130, USA. E-mail: srihari@cs.wustl.edu

  • Venue:
  • Multimedia Tools and Applications
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

The release of DVD-ROMs (Digital Versatile Disk—Read Only Memory), capable of storing an entire 133 minute MPEG-2 movie, coupled withthe gradual deployment of high speed networks will give a much neededimpetus to Video-on-Demand systems, especially movie-on-demandservices. Various memory components such as DVD-ROMs, magnetic RAIDtowers, and RAM are available, each with different storage andbandwidth capabilities, and each with different costs. To design avideo server, we need to optimize the cost factor, yet meet themassive storage, high bandwidth and continuity requirements of videostream delivery.In the first part of this paper we compare DVD-ROM and RAID systemson the basis of two factors—a storage factor and a bandwidthfactor. We work out relations to calculate capacity requirements andcosts, of these systems given the demands of video delivery. We thenpropose an architecture for a video server where we deploy threelayers of memory functioning as video pumps: DVD-ROM towers, RAIDtowers, and RAM. We then work out expressions to determine themigration strategy of a movie between these layers of storage so asto optimize on the cost of storage while satisfying the performancerequirements. We then show the method by which the amount of storagerequired for each type of memory in the three layers can be fixed,knowing the usage pattern.In the second part of the paper, we discuss the implementation of avideo pump where there is a need to integrate scheduling, admissioncontrol, VBR stream management, and handling of VCR-like requestssuch as fast-forward, fast-reverse and pause. We first work outgeneral relations for calculation of the service time of a requestknowing the seek overheads and playback rates. We then show that,surprisingly, the C-SCAN algorithm, which incurs the least amount ofseek overhead and therefore has a lesser cycle time, is twice asefficient as SCAN in terms of buffer space requirements. We thenpropose a ’full-load‘ admission control and scheduling algorithmthat operates on a constant cycle time basis. We test the validity ofthis scheme though a the simulator which we have built based on ourdisk model and use it to extract parameters such as disk bandwidthutilization factor and buffer sizes needed for a VBR load scenario.