Development of energy efficient algorithms for cloud environment

Abstract

In Today s world, Cloud computing has created a revolutionized system for information and newlinecommunication technology. It has enabled computing resource provisioning based on demand newlineservices. The large sets of computing nodes established around the world are connected to large newlinescale data centers with mission critical computing infrastructures. The data centers operate round newlinethe clock to enable the IT industry and transform the way businesses operate. There are two newlinemain factors which are very critical for the data center functioning in the cloud environment. newlineFirst, the ever increasing demand for data computing, processing and storage and Second, the newlinehardware requirements that varies from small to big, run persistently. Data centers consume newlinemassive amount of energy and electricity resulting in huge operating cost and carbon dioxide newline(CO2) emission. As per the recent Gartner report, the Google data center energy consumed newlinethroughout the world is estimated to be 6% - 9% of the entire global electricity usage. This is newlineexpected to grow exponentially, also power consumed in servers through CPU is 46%, memory newlineis 24%, hard disk is 14%, networking is 6%, devices and switches is 8% and others is 2%.Also newlineduring idle mode, data centers consume significant amount of energy. The major challenge of newlinecloud service providers is to minimize the power consumption, adopt methods with a balanced newlinemix of savings with high performance and increase the energy efficiency. newlineThe main objective of the present research work is to develop energy efficient algorithms for newlinecloud environment. The four approaches have been proposed to improve the energy efficiency in newlinethe data centers. In the Method I, an algorithm is developed for the Virtual Machine (VM) newlineselection and its performance on Physical Machine (PM).The shared memory implementation newlinetechniques under different workload conditions were considered. The shared memory newlineimplementation on 9P-virtito file system on the cloud stack platform for different workload newlineconditions is considered.