# 计算机代写|云计算代写cloud computing代考|CSE546

## 计算机代写|云计算代写cloud computing代考|Background

Multi-server energy consumption strategy aims at reducing the power usage of the data center by using the least amount of resources that are available in the PMs while switching off inactive PMs at the data center. Man et al. [9] and Ajiro et al. [10] used different methods, such as “First Fit,” “Best Fit,” to solve VMs allocation problem. A global optimum solution is obtained from the heuristic approaches. “Best fit heuristic solution” is inaccessible because it takes more convergence time, and the proposed work has one more limitation based on a “single objective function.” Beloglazov et al. [11] recommended a “Modified Best Fit Decreasing (MBFD) algorithm” by arranging the ascending order of PMs and descending order of VMs based on their processing capability. In the “First Fit Decreasing (FFD),” VM allocation on PMs is done after VMs and PMs are sorted. The drawbacks are: Distinct allocation goal for VMs and MBFD is not flexible as huge requested VMs get influenced at the data center. The VM allocation issue is solved by using different types of algorithms. Many researchers employed bio-inspired and evolution-inspired algorithms for the cloud data center for allocation of VMs, such as “Genetic Algorithm (GA) [12],” “Particle swarm optimization (PSO),” etc. Xiong et al. [13] used PSO to resolve the issue of allocating energy-efficient VMs. The authors considered a single VM type as it is the major drawback of this research.

“Ant colony”-based VM is proposed for VM allocation by the cloud data centers by Gao et al. [14]. The authors used only a single VM and PM combination which is the drawback of their work. Wang et al. [15] recommended the use of PSO to locate efficient VMs in a data center. The drawback of this research is that the allocations of VMs change the particle velocity that requires more iterations and gives an inaccurate result. “Particle Swarm Optimization (PSO) algorithm” [16] is derived from bird flocking, its a bio-inspired algorithm [17]. It is through PSO that each particle in the swarm gives a solution with four dimensions, its present location, the position, its speed, and the location found among all the particles in the population. Its location is set in the search area based on the position its neighborhood has reached. The limitations of PSO algorithms suffer from the partial optimism that triggers less accurate velocity and distance control. PSO is unable to address particle issues in the field of energy consumption. Sharma et al. proposed “Genetic algorithm and Particle Swarm Optimization (GAPSO)” [6, 8], hybridization of GA and PSO results in increasing the fitness value of the parent chromosomes, and thus allocation of VMs to the PMs is achieved in lesser time.The limitation of GAPSO algorithm takes more convergence time.

## 计算机代写|云计算代写cloud computing代考|Problem Formulation

For each round of scheduling, the essence of the problem is to find a mapping of tasks to worker nodes such that the communicating tasks are packed as compact as possible. In addition, the resource constraints need to be met-the resource requirements of the allocated tasks should not exceed the resource availability in each worker node. Since the compact assignment of tasks also leads to reducing the number of used machines, we model the scheduling problem as a variant of the bin-packing problem and formulate it using the symbols illustrated in Table $5.1$.
In this work, the resource consumptions and availability are examined in two dimensions-CPU and memory. Though memory resources can be intuitively measured in terms of megabytes, the quantification of CPU resources is usually vague and imprecise due to the diversity of $\mathrm{CPU}$ architectures and implementations. Therefore, following the convention in literature [13], we specify the amount of CPU resources with a point-based system, where 100 points are given to represent the full capacity of a Standard Compute Unit (SCU). The concept of SCU is similar to the EC2 Compute Unit (ECU) introduced by Amazon Web Services (AWS). It is then the responsibility of the IaaS provider to define the computing power of an SCU, so that developers can compare the CPU capacity of different instance types with consistency and predictability regardless of the hardware heterogeneity presented in the infrastructure. As a relative measure, the definition of an SCU can be updated through benchmarks and tests after introducing new hardware to the data centre infrastructure.

In this chapter, we assume that the IaaS cloud provider has followed the example of Amazon to create a vCPU as a hyperthread of an Intel Xeon core, ${ }^6$ where 1 SCU is defined as the CPU capacity of a vCPU. Therefore, every single core in the provisioned virtual machine is allocated with 100 points. A multi-core instance can get a capacity of num_of_cores* 100 points, and a task that accounts for $p \%$ CPU usages reported by the monitoring system has a resource demand of $p$ points.

# 云计算代考

## 计算机代写|云计算代写cloud computing代考|Background

.背景信息

Gao等人提出了基于“蚁群”的虚拟机用于云数据中心的虚拟机分配。作者只使用了单一的VM和PM组合，这是他们工作的缺点。Wang等人[15]推荐使用PSO在数据中心中定位高效的vm。该研究的缺点是vm的分配改变了粒子速度，需要更多的迭代，得到的结果不准确。“粒子群优化(PSO)算法”[16]是由蜂群进化而来的一种仿生算法[17]。正是通过粒子群算法，群中的每个粒子给出了一个四维解，包括它的当前位置、位置、速度以及在群体中所有粒子中找到的位置。它的位置是在搜索区域中基于它的邻居已经到达的位置设置的。粒子群算法的局限性在于局部乐观，导致速度和距离控制不够精确。粒子群算法无法解决能量消耗领域的粒子问题。Sharma等人提出了“遗传算法和粒子群优化(GAPSO)”[6,8]， GA和PSO杂交可以提高亲本染色体的适应度值，从而在更短的时间内将vm分配给PMs。GAPSO算法的局限性是需要更多的收敛时间

## 计算机代写|云计算代写cloud computing代考|Problem Formulation

.问题表述

myassignments-help数学代考价格说明

1、客户需提供物理代考的网址，相关账户，以及课程名称，Textbook等相关资料~客服会根据作业数量和持续时间给您定价~使收费透明，让您清楚的知道您的钱花在什么地方。

2、数学代写一般每篇报价约为600—1000rmb，费用根据持续时间、周作业量、成绩要求有所浮动(持续时间越长约便宜、周作业量越多约贵、成绩要求越高越贵)，报价后价格觉得合适，可以先付一周的款，我们帮你试做，满意后再继续，遇到Fail全额退款。

3、myassignments-help公司所有MATH作业代写服务支持付半款，全款，周付款，周付款一方面方便大家查阅自己的分数，一方面也方便大家资金周转，注意:每周固定周一时先预付下周的定金，不付定金不予继续做。物理代写一次性付清打9.5折。

Math作业代写、数学代写常见问题

myassignments-help擅长领域包含但不是全部: