My topic is cloud computing.my thesis topic is enhancing the performance of the task scheduling for virtual machines in the cloud environment.for this, I need to develop to round-robin algorithm. in this algorithm need to be completely standard. if it is not in standard my guide will be caught within 2 minutes. algorithm needs to go through all the test cases and I required the calculations of the time complexity. I am attaching my dissertation paper u will get some idea what u need to do? in this attached pdf I proposed one technique for round-robin but my thesis guide reject my idea.
Enhancing performance of Task Scheduler on Virtual Machines in Cloud Environment Narendra Jasti x18170374 MSc in Cloud Computing 29th July 2019 Abstract Cloud computing plays a very significant role in information technology services. Cloud provides a variety of services that can be accessed either remotely or virtual resources. Due to the rapid usage of cloud computing, there is a rise in problems. Among those problems, task scheduling is one of the critical problem. It does receive a task from the users and forwarded to the virtual machine (VM) with the objective of less waiting time and complete resource usage. The approach of this research is to provide the best solution for task scheduling in the cloud by using a proposed round-robin algorithm. The proposed algorithm helps to reduce not only waiting time but also the overall execution time of the tasks and resource utilization. Round robin algorithm has some unique features such a simple calculation rule, less starvation and act also dynamically depends on cloud situations and applicable for the load balancing. Keywords: Cloud Computing, Virtual Machines, Round Robin, Task Scheduling, Dynamic Time Quantum. Contents 1 Introduction 2 2 Research Question 3 3 Literature Review 3 3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 3.2 Existing Task Scheduling Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 3.3 Comparison of various existing algorithms in task scheduling . . . . . . . . . . . . 9 4 Proposed Approach 9 5 Proposed Methodology 11 6 Proposed Implementation 13 7 Proposed Evaluation 14 8 Planning 17 References 17 1 1 Introduction According to the National Institute of science technology Cloud computing is defined "as a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provi- sioned and released with minimal management effort or service provider interaction." In cloud computing, we can scale Infrastructure, which is linked with the specified number of systems. Clients can utilize many services like storage, networking, and virtual servers. These services are accessible through the internet. The most significant advantage from the cloud is that it can provide anything as a service (Xaas) which means you can use services like Infrastructure as a service, platform as a service and software as a service. Even though cloud computing is providing different types of services, but there are a different number of problems in cloud computing. In those problems, task scheduling is one of the most critical problems. "Task Scheduling problems are concerned with searching for optimal (or near-optimal) real-time and predictive schedules subject to several constraints"[1]. Task scheduling plays a vital role in cloud computing. Task scheduling works on mapping technique from client’s tasks to opting for resources and their execution. Cloud computing has completely new and unique features when compared with grid computing in terms of task scheduling. Cloud computing enables virtualization and flexibility features. With the help of virtualization technology, all physical resources like storage, memory, and CPUs are virtualized and ultimately allow users to utilize these services. Every user has a unique virtual device; these devices do not communicate with each other and are created based on client demand [2]. Each virtual machine can decrease the measure of physical servers, and increases the usage of server resources. Each physical server has its own set of resources. Flexibility in resources is provided by taking advantage of the cloud, which can be automatically scaled up and scale down according to user requirements for tasks by having these unique sets of features. Task scheduling mechanism is not suitable for grid computing [2],[3]. Figure 1: Scheduling process in cloud The rapid development of cloud computing has faced many problems and challenges, such as resource management, security, and performance. Task scheduling is wholly related to resource management and has a massive impact on resource usage, throughput, and efficiency. Cloud task scheduling deals with allotting user’s tasks to resources with enhanced throughput and system usage without disrupting SLA demand. The optimization problem is correct task scheduling and mapping user requests to resources. Even scheduler seeks to detect the best optimal virtual machine mapping concerning scheduling times such as an execution time and response time. Consider there are only n tasks T= T1, T2, T3,.... TN and M resources R= R1, R2,R3.... RM in the present system of computing. Assigning cloud resources to virtual resources is done by 2 direct mapping of individual resources [4]. The primary focus is to decrease waiting times, and total processing that is related to scheduling with increased system throughput and quality of service (QoS). Generally, the process of issuing virtual machines among a different number of tasks is to decrease the workload.There are certain parameters that are suspectable to change such as turnaround time, power utilization and response time can be taken into consider. On the other hand, task scheduling is treated as a "Non-polynomial" (NP)-hard problem. Accordingly, there is an urgent requirement for proper optimization for solving polynomial time [5]. In cloud computing task scheduling is the challenging issue. There is a scope in improving task scheduling performance like cost for execution time and Quality of service(Qos). Advantages of both the clients and service providers can also be taken into consideration. The main agenda of this research proposal is to improve the performance of task scheduling parameters like execution time, turnaround time, and waiting time. The remaining part of this paper is as follows. Next section 2 consists of Research Question, section 3 consists of Literature Survey about different issues with task Scheduling and task scheduling algorithms of the virtual machine are discussed, Section 4, 5 6 and 7 consists of the Methodologies, Implementation and Evaluation details about the Proposed approach. 2 Research Question 1. Does improving the round-robin algorithm will perform the best execution for task scheduling in cloud computing? 2. Can improvisation of turnaround time and weighting time optimize task-based scheduling on the basis of extended Round Robin algorithm? 3 Literature Review In this section, a brief review study is presented which shows most relevant research work done for enhancing the performance of task scheduling in the cloud. This review includes various algorithmic solutions that are presented. 3.1 Background In the year 2010[6], proposed that a hybrid solution that can deliever optimal task Scheduling performance. Generally, the combination of both private and public cloud is known as a hybrid cloud. The problem scenario here is that workloads have to be outsourced from the private cloud to Public Cloud during peak workload intervals if there are not sufficient resources available in the private cloud. These workloads are to be constrained by the deadline and Quality of Service (QOS) requirements. In this case, the Decision making process is required to select different kinds of workloads that need to be outsourced and later forwarded to the selected cloud provider. In this way, the utilization and cost of running outsourcing tasks are low in data centers. A linear programming technique is used to tackle this problem, and it outperformed well in terms of cost minimization, feasibility, and scalability. In 2011, two below-mentioned algorithms are proposed for task Scheduling based on the processing of tasks and computational capacity of resources by Sindhu, S. and Saswati Mukher- jee [7]. First, an algorithm named as Longest Cloudlet Fastest Processing Element (LCFPE) 3 this algorithm looks at very high computational complexity of tasks in the process of making Scheduling decisions. It sorts tasks in the descending order of length and sorts processor in high priority of processing power. Then maps both cloudlets and Processor on one-to-one mapping which tries to reduce makespan. Second algorithm is known as Shortest Cloudlet Fastest Pro- cessing Element (SCFP). It does the opposite of Longest Cloudlet Fastest Processing Element (LCFPE). It sorts tasks in the smaller size of length and processors in descending order then map jobs got settled from one sorted list to another sorted list of processors based on the one-to-one mapping basis. This algorithm reduces waiting time for longer jobs. They also suggested for future heuristic methods to consider the priority of tasks. Ant Colony Optimization Algorithm was initially proposed by Marco Dorigo in (1992). It is based on the analogy of real ants. Ants can find an optimal path from its nest to food and from food sources taken back to the nest. Ant follows the same optimal path by its released secretion pheromone. In the case of task scheduling, tasks are analogous to ants, and Virtual machines imitate food resources. Ant Colony Optimization can be used to explain many combinative optimized problems with different targets of costs and performances. A complexity of Ant Colony Optimization (ACO) can be measured in the cloud with the help of several available R resources T tasks and N number of tasks. These can be classified in time and space complexity. "So, finally time complexity of Ant Colony Optimization is O(TN), and space complexity is O (1), Since this algorithm consumed with an endless number of tasks and resources, but it does not include any dynamic variables"[8]. Gogulan, Kavitha and Karthik Kumar in 2012 [9] started a new algorithm called "Multiple Pheromone Algorithm" (MPA). The main intention of "multiple pheromone algorithm" is to make cloud-related services more dynamic and resource provisioning. In cloud computing, one of the significant problems is job scheduling because it increases the weight of the grid and becomes too challenging to solve correctly. "Multiple Pheromone Algorithm is belonging to ant colony optimization". The central theme of the "Multiple Pheromone Algorithm" generates a proper schedule for finishing the task in less period and using the resources without usage. In this paper, the author compared three different parameters like "cost, quality of service and reliability". In this comparison, the "Multiple Pheromone Algorithm" provides the best solution when compared with Ant colony optimization. In the year 2013, there was a situation where a certain number of tasks were not coming in correct order for the processing of tasks to the cloud. Because of this problem, outcomes are slow in assigning tasks to virtual machines. For skipping this, the author suggests that the genetic algorithm is the most