Design of a Fault-Tolerant Distributed Storage System via Non-Systematic Fountain Codes

International Journal of Computer Science and Technology | Vol. 13, No. 11, November 2011 | pp. 71–77
DOI: 10.46882/2011/IJCST/000117

Research Article

Title: Design of a Fault-Tolerant Distributed Storage System via Non-Systematic Fountain Codes

Names of Authors: David R. Sterling¹, Marie S. Curie²

Authors’ Affiliations: ¹Department of Computer Science, Stanford University, Palo Alto, USA; ²Laboratory of Information Systems, Pierre and Marie Curie University, Paris, France

Abstract: Distributed data systems deployed over commodity computing clusters frequently experience node dropouts, corrupt files, and network disconnections. Standard data duplication techniques minimize packet loss risks but incur large storage space penalties across modern enterprise storage centers. This study details the design and deployment of a Fault-Tolerant Distributed Storage System (FTDSS) that uses non-systematic Fountain codes, specifically Luby Transform (LT) codes, to protect data files efficiently. The system fragments incoming files into equal blocks and produces an open-ended stream of encoded packets. Storage nodes accept these packets asynchronously, and original files can be fully recovered by any node that collects a slight overhead of symbols (surplus factor <= 1.05). Tests across an 8-node lab testbed demonstrated that FTDSS tolerates a concurrent node failure rate of 37.5% without incurring any data loss. Storage efficiency metrics surpassed traditional replication models by 2.4 times, while maintaining read/write throughput speeds at 85 MB/s. This coding strategy offers a reliable, low-overhead solution for cloud archives and enterprise backup infrastructures.

Keywords: Distributed Storage, Erasure Coding, Fountain Codes, Fault Tolerance, Data Recovery, Storage Efficiency

Manuscript Timeline: Received August 04, 2011; Revised September 18, 2011; Accepted October 05, 2011; Published November 20, 2011.

Citation: Sterling, R. D., & Curie, M. S. (2011). Design of a Fault-Tolerant Distributed Storage System via Non-Systematic Fountain Codes. International Journal of Computer Science and Technology, 13(11), 71–77.

Subscribe to read the full article: https://www.internationalscholarsjournals.org/subscribe-to-read