Design and Evaluation of Ring-Based Video Servers
Video-on-demand servers (VODS) must be based on cost-effective architectures. Therefore architectures based on clusters of PCs will probably be the most suitable to build VODS; they provide the same performance of a large server through the aggregation of many smaller, inexpensive nodes. In this paper, we show that the interconnection network used to build the cluster can be very expensive if it is based on a set of switches. To obtain the most cost-effective architecture, we argue that VODS must be Dual Counter-Rotating Ring-based (DCRR-based). DCRR are very inexpensive and fulfill all the basic criteria needed for VODS architectures except scalability. To address the scalability issue, we propose to enhance the design of DCRR by partitioning it logically in combination with three new policies (“fast stream migration”, “stream splitting” and “distributed XORing”). This design brings very cost-effective and scalable VODS able to play up to 13500 MPEG-2 concurrent streams using 252 nodes.
KeywordsInterconnection Network Storage Node Load Imbalance Spatial Reuse Delivery Network
Unable to display preview. Download preview PDF.
- 1.ANSI X3T10.1/0989D revision 1.0, “Information Technology-Serial Storage Architecture-Transport Layer 1 (SSA-TL1)”, American National Standard Institute April 1996.Google Scholar
- 2.Bach C. and Grebe A., “Comparison and performance evaluation of CRMA-II and ATMR”, Eleventh Annual Conference on European Fibre Optic Communications and Networks, EFOC&N’ 93, June 1993.Google Scholar
- 3.Shenoy P.J. and Vin H.M., “Efficient Striping Techniques for Multimedia File Servers”, in Proceedings of the Seventh IEEE International Workshop on Network and Operating System Support for Digital Audio and Video (NOSSDAV’97), pp. 25–36, May 1997.Google Scholar
- 4.Chervenak A.L., “Tertiary Storage: An Evaluation of New Applications”, Phd Thesis, University of California at Berkeley, 1994.Google Scholar
- 5.Cidon I. and Ofek Y., “MetaRing-A Full-Duplex Ring with Fairness and Spatial Reuse”, IEEE Transactions on Communications, Vol.41, No.l, Jan 1993.Google Scholar
- 7.Tetzlaff W. and Flynn R., “Block Allocation in Video Servers for Availability and Throughput”, Multimedia Computing and Networking, 1996.Google Scholar
- 8.Dan A. and Sitaram A., “An Online Video Placement Policy based on Bandwidth to Space Ratio (BSR)”, Proceedings of the ACM SIGMOD International Conference on Management of Data, pp. 376–385, San Jose, May 1995.Google Scholar
- 9.Lee E.K. and Katz R.H., “Performance Consequences of Parity Placement in Disk Arrays”, Proceedings of the Fourth International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS-IV, pp. 190–199, April 1991.Google Scholar
- 10.Shenoy P.J. and Vin H.M., “Failure Recovery Algorithms for Multi-Disk Multimedia Servers”, Univ. of Texas at Austin, Department of Computer Sciences, Technical Report 96-06, April 1996.Google Scholar
- 11.Tewari R., Mukherjee R. and Dias D.M., “Real-Time Issues for Clustered Multimedia Servers”, IBM Research Report, RC20020, June 1995.Google Scholar
- 13.“SRC3266DE Preliminary Data Sheet, Sebring Ring Connection for PCI 32”, http://www.sebringring.com, 1999.
- 14.Varma A. and Raghavendra C.S., “Interconnection Networks for Multiprocessors and Multicomputer-Theory and Practice”, IEEE Computer Society Press, 1994.Google Scholar