# 计算机代写|云计算代写cloud computing代考|CSE356

## 计算机代写|云计算代写cloud computing代考|Parametric Hypothesis Testing and Test Error

Results for normality testing through the S-W normality test have suggested that elements of $\bar{E}^R$ and $\bar{E}^S$ follow normal distribution, which meets the prior condition of parametric tests. Positive numeric value of every element of $X^{\bar{E}}(3.12)$ articulated in Table $3.4$ refers to the fact that energy consumption by SRTDVMC is numerically lower compared to $R T D V M C$ for different workload scenarios as presented in experiments. We hence aim to perform parametric hypothesis test to find whether simulation output samples, $X_{N T, P L}^{\bar{E}}$ (3.12) featuring difference between two corresponding means of $R T D V M C$ and SRTDVMC associated to a unique combination of Nectar and PlanetLab workload is statistically significant. Our sample size is less than 30 and means of no more than two DVMC algorithms (i.e., SRTDVMC and $R T D V M C$ ) would be compared. Therefore, among different parametric tests we opt to use the $t$-test, instead of $Z$-test, $F$-test, and ANOVA. Based on the data samples, the $t$-tests can be classified into three groups: One sample, Two Independent Samples and Paired Samples t-test. For a specific combination of $N L$ and $P L$, corresponding $\bar{E}{N T, P L}^R$ and $\bar{E}{N T, P L}^S$ has a relationship, as $\bar{E}{N T, P L}^R$ and $\bar{E}{N T, P L}^S$ represent $\bar{E}_{\mathrm{CDC}}$ for $R T D V M C$ and SRTDVMC respectively, under a particular workload distribution scenario. Therefore, the paired two tail $t$-test is performed.

The null hypothesis with the $t$-test is that mean $\mathrm{CDC}$ energy consumption with RTDVMC, $\bar{E}^R$ and mean CDC energy consumption with SRTDVMC, $\bar{E}^S$ are same, while the alternative hypothesis is that $\bar{E}^S<\bar{E}^R$. Utilizing (3.17)-(3.19), the test statistic for the $t$-test, denoted by $t_{\bar{X}} \bar{E}$ is found as $2.13$ and the corresponding $p$ value is found as $7.10693 \times 10^{-6}$, which is lower than critical value, $\alpha$ as $0.05$. For clear understanding of the interpretation of the $t$-test result, we have first explained the performed the $t$-test in more details in the following.

## 计算机代写|云计算代写cloud computing代考|Conclusions and Future Work

While correlation exists between VMRT and energy consumption, traditional DVMC algorithms except RTDVMC do not consider heterogenous VMRT in VM consolidation decision process. Furthermore, existing algorithms consolidate VMs in as few PMs as possible based on the premise that optimal energy efficiency can be achieved with maximum load on PM. However, for state-of-the-art PMs, energy efficiency rather drops above $70 \%$ load level. Combining lack of consideration of heterogeneous VMRT and ignoring changed energy-efficient characteristics of underlying PMs, existing DVMC algorithms lack in performance in the context of real Cloud scenario with heterogeneous VMRT and state-of-the-art PMs.

RTDVMC considers heterogeneous VMRT. However, issues with RTDVMC are twofold – first, it does not take the changed energy-efficiency characteristics of modern PMs into account and second, it only aims to minimize energy consumption without considering VM migration minimization. VM migration, nonetheless, increases network overload causing degraded QoS and increased energy consumption by networking equipment. VM migration being an unavoidable part of VM consolidation, minimizing both energy consumption and VM migrations at the same time are confronting objectives. As such, in this paper, we have brought forth a novel multi-objective DVMC algorithm, namely, $S R T D V M C$, which aims to reduce VM migrations without compromising energy efficiency. Consideration of heterogeneous VMRT values in VM consolidation decision process enables SRTDVMC to be more energy efficient. On top of that, contrast to RTDVMC, SRTDVMC incorporates consideration both benefit and cost prior any VM migration. As a result, it is robust against the changed energy-efficiency characteristics of underlying PMs and can reduce. VM migration without compromising energy-efficiency compared to RTDVMC.

Performance of SRTDVMC has been tested through the most popular Cloudbased simulation tool, namely, CloudSim, in the context of hundreds of different cutting-edge PMs and thousands of VMs representing heterogenous VMRT of real Nectar Cloud, as the assigned workload reflects real Cloud workload obtained from PlanetLab. The empirical outcome exhibits the superiority of SRTDVMC over RTDVMC in both metrics – CDC energy consumption and VM migration. Three key elements are extracted from our research. First, based on our experiments, VMRT impacts on both aspects – energy consumption and VM migration, and hence, DVMC algorithms are needed to be developed considering the presence of heterogeneous VMRT. Second, such working principal of existing algorithms that maximum energy efficiency is achievable at maximum load on PM is found as false for state-of-the-art PMs, resulting into performance inefficiencies. Our proposed SRTDVMC algorithm addresses this issue. Third, simulation results show that if corresponding cost and benefit are considered prior VM migration, then concomitant optimization of both aspects – reduction of energy consumption and VM migration can be achieved. In the following section, we have suggested several future research pathways to further improve the energy-efficient management of CDC.

# 云计算代考

## 计算机代写|云计算代写cloud computing代考|参数假设检验和测试误差

. .计算机代写|云计算代写cloud computing代考|

## 计算机代写|云计算代写cloud computing代考|结论和未来的工作

.

RTDVMC考虑异构VMRT。然而，RTDVMC的问题有两个方面——首先，它没有考虑到现代pm的能源效率特征的变化;其次，它只致力于最小化能源消耗，而不考虑VM迁移最小化。虚拟机迁移会增加网络过载，导致QoS降低，网络设备能耗增加。VM迁移是VM整合中不可避免的一部分，同时最小化能耗和VM迁移是我们面临的目标。因此，在本文中，我们提出了一种新的多目标DVMC算法，即$S R T D V M C$，其目的是在不影响能源效率的情况下减少VM迁移。在虚拟机整合决策过程中考虑异构VMRT值可以使SRTDVMC更节能。最重要的是，与RTDVMC相比，SRTDVMC在任何VM迁移之前都考虑了收益和成本。因此，它对潜在pm的能源效率特性的变化具有鲁棒性，并且可以降低。与RTDVMC相比，VM迁移不会影响能源效率

SRTDVMC的性能已经通过最流行的基于云的仿真工具CloudSim进行了测试，该工具在数百个不同的前沿pm和数千个vm的背景下表示真实甘露云的异构VMRT，因为分配的工作负载反映了从PlanetLab获得的真实云工作负载。实证结果表明SRTDVMC在CDC能耗和VM迁移两个指标上优于RTDVMC。从我们的研究中提炼出三个关键要素。首先，根据我们的实验，VMRT对能耗和VM迁移两个方面都有影响，因此需要考虑异构VMRT的存在，开发DVMC算法。其次，现有算法认为在PM负载最大的情况下可获得最大能源效率的工作原理对于最先进的PM是错误的，从而导致性能效率低下。我们提出的SRTDVMC算法解决了这个问题。第三，仿真结果表明，如果在虚拟机迁移之前考虑相应的成本和收益，则可以实现降低能耗和虚拟机迁移这两个方面的同时优化。在接下来的部分中，我们提出了未来进一步完善CDC节能管理的几条研究路径

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: