Experimental Study on Communication Overhead of Data Processing in Distributed System
Date
2019
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Imam Abdulrahman Bin Faisal University
Abstract
The society is faced with the biggest growth in computing for the past two decades. With high levels amounting up to 2.5 exabytes being generated on a daily basis, it is essential to have communication networks that allow for the seamless exchange of data while ensuring that the processes are both secure and, therefore, reliable. This thesis explores experimental study on communication overhead of data processing in distributed system. Distributed systems carry with them substantial needs. As will be explored in this thesis, elements of scalability and remaining concurrent are key features that are the pillars of any distributed system. This thesis seeks to illustrate the characteristics, as mentioned above, reflection on contextual computing settings. Through the "Lazy Execution on Demand Serve" (LEDs) algorithm, the thesis will illustrate the working capabilities against existent algorithms. Having various weaknesses, the existing algorithms have some weak areas, which the researcher felt worth detailed scrutiny. Working on a high leveled architectural framework, the algorithm presents a description of an algorithm that is both seamless and also secure. Additionally, the algorithm's scheduling functions are more streamlined in relation to the compared algorithms. The thesis focus is to test claims; claims of the efficiency of system’s ability to schedule operations, a feature which the thesis accomplishes.
Description
Keywords
Socket Programming, Communication Overheads, Virtual Machines, Distributed Systems, Round Trip Time