Scalability of a Workstation Cluster Architecture for Video-on-Demand Applications

  • Authors:
  • Peter H. Hughes;Gunnar Brataas

  • Affiliations:
  • -;-

  • Venue:
  • TOOLS '00 Proceedings of the 11th International Conference on Computer Performance Evaluation: Modelling Techniques and Tools
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

Video-on-demand (VoD) applications require architectures which are scalable over a wide range of capacity. Two of the most promising architectures for VoD are massively parallel symmetrical computers based on distributed memory (MPP) and clusters of workstations interconnected by a high-performance network. The work described here investigates the scalability of a prototype cluster architecture and compares it with a commercially available hypercube MPP system. Measurement experiments using emulated workloads were carried out for increasing loads, and the effects on capacity of encoding-mode and user interactions were explored. Static analysis, supported by simulation results was used to make projections over a range of configurations and technology assumptions. Sample results are presented which chart scalability on two hierarchic levels.