Modeling and Analysis of Multilingual Information Parallel Downloads in Data Grid

The need arises in parallel downloads of multilingual information for powerful graphical and analytical tools, as information with a variety of different languages distributed in different Web pages and the databases are heterogeneous and uneven in data grid. Petri net is a powerful graphical and mathematics tool for describing the concurrent, asynchronous and dynamic events. The parallel downloading of multilingual information was modeled and analyzed using extended timed colored Petri net (ETSdCPN). In ETSdCPN model, the color represents different languages information, and the time duration associated with place instead of transition is a function of tokens instead of constant. The reachable parallel download graph (RPDG) of ETSdCPN is defined. Finally, some important results such as rate of satisfaction and makespan of multilingual information parallel downloads are gotten by analyzing reachability of Petri net.

Mechanisms for Parallel Data Transport

Evolving paradigms of parallel transport mechanisms are necessary to satisfy the ever increasing need of high performing communication systems. Parallel transport mechanisms can be described as a technique to send several data simultaneously using several parallel channels. The authors’ survey captures the entire building blocks in designing next generation parallel transport mechanisms by firstly analyzing the basic structure of a transport mechanism using a point to point scenario. They then proceed to segment parallel transport into four categories and describe some of the most sophisticated technologies such as Multipath under Point to Point, Multicast under Point to Multipoint, Parallel downloading under Multipoint to Point, and Peer to Peer streaming under Multipoint to Multipoint. The Survey enables the authors to stipulate that high performing parallel transport mechanisms can be achieved by integrating the most efficient technologies under these categories, while using the most efficient underlying Point to Point transport protocols.

A Multi-Node Co-Allocation Parallel Downloading Algorithm

In distributed data storage, a particular dataset can reside at multiple locations in order to get high availability .Thus, the dataset can be downloaded in parallel from multiple nodes.Throughput between server and client changes dynamically, so the downloading speed can vary unpredictably. A dynamic parallel downloading algorithm based on measurement of bandwidth and bandwidth prediction is produced in this paper and server caching is adopted in order to improve downloading speed. The algorithm dynamically adjusts downloading of the last block to make parallel downloading from multiple servers end almost simultaneously. With this approach, the download time is reduced and the robustness of the downloading system is improved. Besides, the algorithm not only avoids complicated server selecting mechanism, but also improves load balance of the servers.