Mitra Project Information Page
MITRA
A C
ONTINUOUS MEDIA SERVER


USC Database Lab

PROJECT DESCRIPTION

The research activities at the USC Database Lab during the past three years have resulted in the design and implementation of Mitra. Mitra is a scalable continuous media server that is operational on a cluster of HP workstations (model 9000 735/125). It supports a hierarchical storage structure by treating the available disk space as a cache. Several on-going projects are investigating applications of Mitra to different domains, e.g., sound-servers, non-linear digital editing, etc. In the future, we envision extensions of Mitra in support of both traditional data as well as continuous media.

Mitra is a software based system that employs off-the-shelf hardware components. Its present hardware platform is a cluster of multi-disk workstations, connected using an ATM switch.


HIGHLIGHTS

Commercial, off-the-shelf hardware components (workstations, networking)
Scalable, multi-workstation implementation
Support for a mix of media types: audio, video
Support for multi-speed fast-forward and fast-rewind for video streams (see slides presented at WITS '97 workshop)


RESEARCH TEAM

Dr. Shahram Ghandeharizadeh
Roger Zimmermann
Weifeng Shi
Jaber Al-Marri
Seon Ho Kim
Reza Rejaie
Dr. Douglas J. Ierardi
David Ta-Wei Li
Weifang Xie


ONLINE PAPER

An overview of Mitra was published as chapter three in the book Multimedia Technologies and Applications for the 21st Century (Editor Borko Furht, Kluwer Academic Publishers, Boston, December 1997, ISBN: 0-7923-8074-6). The article first appeared in the Multimedia Tools and Applications journal (Kluwer Academic Publishers, 5(1):79-108, July 1997). Here is an online version of this paper (created with Hyperlatex 1.4, Copyright © 1994 Otfried Schwarzkopf).

PostScript Version of Paper (187 KB, zipped)
PostScript Version of Paper (311 KB, compressed)
Photograph (117 KB)  ||  Related Papers  ||  Developer's Page

Please send comments, questions, or any feedback to rzimmerm@imsc.usc.edu.
This research was supported in part by Hewlett-Packard with an unrestricted cash/equipment gift, and the National Science Foundation under grants IRI-9203389, IRI-9258362 (NYI award), and CDA-9216321.


Last updated: Wednesday December 2, 1998.
All Rights Reserved © USC Database Lab 1996-2002