A scheme for high-performance data delivery in the Web environment

Abstract

The paper describes a scheme for high performance and dependable data storage and delivery in a large scale distributed computing and communication environment such as the Web environment. The proposed scheme utilizes the parallelism of several distributed data servers storing striped data blocks to achieve high throughput. It employs coding techniques to protect the system against data unavailability and hence achieve dependable service. The performance results show that the proposed method has several advantages over traditional ones, such as data service through mirror sites. The error probability of the proposed method is orders of magnitude smaller than that of the mirroring with the same redundancy rate. The data rates for file downloading could be improved significantly by the proposed scheme.