QOS Enhanced Energy Aware Task Scheduling Models in Cloud Computing

Abstract

The Cloud computing is the on-demand availability of a resource that is delivered to us as-a-service and all data served are stored in data centers. So, all the Cloud service providers-built data centers to host the applications. But, to run these data centers there is a requirement for more energy and thus emits large carbon. Hence, it is required to optimize the energy consumption at data centers, by understanding the energy flow and how is it distributed. The energy consumption of data center can be reduced through resource allocation and task scheduling. The existing methods are not sufficient for resource management in Cloud, it is because they are not intelligent enough to allocate the resources to the given application as per the QoS requirements of the user. The research work presented in this chapter presents four new contributions to optimize energy consumption. In order to solve the scheduling issue, the energy aware multi-objective genetic algorithm for Task Scheduling (MOTS) takes into account energy usage, makespan, and data transfer time. By utilizing the Ant Colony Optimization (OSACO) algorithm, an optimized scheduling mechanism is created, which lessens the demands on the system's resources and saves money and time. Assesses it in terms of makespan, deadline violations, and cost. In order to optimize for a number of competing goals, like energy consumption, data transfer time, and makespan, the energy aware multi-objective genetic algorithm for task scheduling has been proposed as an improvement over pre-existing algorithms. Users are given the freedom to choose among the goals and schedule their tasks accordingly. Cost-based model is for resource selection and an optimization technique for resource scheduling is proposed which negotiates the cost between both the parties, client and Cloud service provider. The resources are chosen using the suggested auction methodology and the offer made by the Cloud service provider. Time and money constraints are taken into account when using an ant colony optimization mechanism to assign work to available resources. In order to meet the needs of the users, a multi-objective algorithm is suggested that uses the non-dominated approach and the crowding distance technique to calculate the quality of service for the virtual machines before allocating them to jobs. Makespan, deadline violations, and cost are used to gauge the effectiveness of the proposed multi-objective model in comparison to previous models like MOTS and OSACO algorithms. In respect of energy consumption, makespan, and transmission time of data, multi-objective genetic algorithm for task scheduling outperformed the state-of-the-art methods by a margin of 20.4%, 23.9%, and 12.5%, respectively, for the objectives that were prioritized. Money and time are not factors in the MOTS approach. In OSACO, in addition to the energy consumption the cost and time are also considered. There is a wide range of values for the OSACO's energy usage, from 5 J for 2 resources to 36 J for 10 resources. Though the proposed OSACO approach has a mechanism for efficiently matching tasks and resources based on the needs of the client's quality of service, it falls short when it comes to optimizing the cost in a way that is mutually beneficial to the user and the provider in terms of the time it takes to complete the tasks. The proposed ACO auction model minimizes the cost in selecting the resources and improving the resource allocation. The cost spent $124 was very less amount, completion time taken was 1.21 × 103 and recorded 24 J, 18 J, and 12 J for 90, 60, and 30 task arrival rates and is noted that, it is efficient in minimizing the completion time, cost and reducing the energy consumption but still it is required to focus on deadline violations. When compared to MOTS and OSACO, the multi-objective dynamic resource scheduling model's makespan decreases by 23% and 19%, respectively, for workloads of 1000 jobs. In comparison to existing algorithms, the proposed one reduces deadline violations by 18% and saves an additional $120 in costs.