当前位置:首页 >> 英语学习 >>





论文题目: A Data-Placement Strategy Based on Genetic Algorithm in Cloud Computing 所选部分: Introduction Summary and Future Work 1.1.2

论文原文内容 Introduction With the increase of network equipment as well as the development of the Internet, data generation and storage capacity are growing explosively(爆发的,引爆的); data centers will face unpredictable visitor volume (不可预知的访问者数量) . The large amount of data and the complex data structures make traditional database management unable to meet the requirements of big data storage and management. The distributed architecture(分布式体系结构) of cloud computing can provide high-performance computing resources and mass storage resources. However, in distributed cloud computing system, data-intensive (数据密集型) computing needs to deal with large amounts of data; in multi-data center environment, some data must be placed in a specified data center (规定 的数据中心) and cannot be moved. A computation may process datasets from different data centers, then data scheduling between data centers will occur inevitably (不可避免的) . Because of the huge size of data and limited network bandwidth, data scheduling between data centers has become a huge problem. The datasets processed simultaneously (同时的) by a computation should be placed in the same data center, then almost all data processing is completed locally; that is the basic idea of the paper. Much work has been developed about the data placement in distributed system and they can be divided into two types in general: static data placement and dynamic data placement. Most static data placement algorithms require complete knowledge of the workload statistics such as service times and access rates of all the files. Dynamic data placement algorithms, generate file-disk allocation schemes on-line to adapt to varying workload patterns without a prior knowledge(先前 知识) of the files to be assigned in the future. Dynamic data placement strategies update the placement strategy potentially upon every request. Obviously, they are effective when the data size is relatively small such as the case in web proxy caching (网络代理缓存) . However, in applications like distributed video servers(分布式视频服务器), dynamic schemes become less useful. In data-intensive computing, if multiple computations jointly process multiple datasets in a frequent way, these datasets are supposed to be correlative with (与什么有关联) each other. Some researches on data placement are based on data correlation; however, the definitions of data correlation are not reasonable, and no effective method is proposed to reduce the data scheduling between the data centers. Replica strategy (复制策略) is an effective measure to reduce the data scheduling and has earned widespread research interests, and it is also based on data placement. This paper presents (提出) a genetic algorithm-based data placement strategy. First, a mathematical model of data scheduling between the data centers in cloud computing is built, and
第 1 页


the fitness function(适应度函数) based on the objective function(目标函数) is defined to evaluate the fitness of each individual in a population. After the initial population generated in accordance with the principle of(以什么原则) survival of the fittest, the evolution of each generation produces better approximate solution. In every generation, roulette-wheel selection; (轮盘赌选择) is used to choose the appropriate individuals with high fitness value and the individuals with low fitness value are eliminated. With the crossover and mutation operations, we change the placement location of datasets. Under the principle of survival of the fittest (适者生存) , the optimal individual can be found during the evolution. Summary and Future Work In the environment of distributed cloud computing, placing data to the appropriate data center has become a critical issue(一个关键问题). Reasonable placement of datasets in data centers can minimize the number of data scheduling between the data centers. In this paper, a mathematical model is built to illustrate(说明) the relationship among datasets, data centers and computations. Three different algorithms are used to search the approximate optimal data placement matrices(矩 阵; 模型) . By comparing genetic algorithm with exhaustive search algorithm (穷举搜索算法) and the Monte Carlo algorithm, we can work out the truth that under verifiable(可证实,可检验) conditions, genetic algorithm can find the optimal data placement matrix; when the number of datasets is large enough, genetic algorithm can find an approximate optimal data placement matrix in a reasonable time, and the optimization result is better than Monte Carlo algorithm. Currently, the focus of our research is to find an optimal data placement matrix, making the number of data scheduling between the data centers as small as possible. During the research, the impact of data access history and access heat on data placement are out of our consideration(出乎我们的考虑). The heat of the data and the execution frequency of computations are not constant over time, then data placement needs to update which increases the cost of data management for enterprise; this issue needs further study. In terms of(在什么方面) genetic algorithms, the selection is an important operator. There are many selection methods, such as Roulette wheel selection method, league selection method, expectations selection method. In this paper we use Roulette wheel selection method. Different methods of genetic selection affect the performance of the algorithm which requires further study. 1.1.3


词汇习得:相关生词已经在文章中标注。 1:unpredictable adj. 不可预知的;不定的;出乎意料的 n. 不可预言的事 eg: In absence of comprehensive information on how the software works, the user will have an impression that the its behavior is unpredictable. 如果没有面面俱到的信息来描述软件的工作方式,那么用户就会觉得它的行为是不可预 知的。 2:specified adj. 规定的;详细说明的
第 2 页

v. 指定;详细说明(specify 的过去分词)


eg:The fund was specified to maintain the ancient buildings. 该项基金被指定为专门从事维修古建筑而使用。 3:exhaustive adj. 详尽的;彻底的;消耗的 eg: These reasonsare by no means an exhaustive one, but I think they are among the mostimportant reasons. 虽然这些原因不会是全面详尽的,但是我认为它们是最重要的原因。 句子习得: 1:With the increase of network equipment as well as the development of the Internet, data generation and storage capacity are growing explosively(爆发的,引爆的); data centers will face unpredictable visitor volume(不可预知的访问者数量). 分析:这个句子中 with 引导的是一个伴随状语,可以翻译成随着什么。As well as 翻译成和。 face 这里做动词,面对的意思。 2:In data-intensive computing, if multiple computations jointly process multiple datasets in a frequent way, these datasets are supposed to be correlative with(与什么有关联)each other. 分析:be supposed to 被认为怎么样,be correlative with 与什么有关联。 3:This paper presents(提出) a genetic algorithm-based data placement strategy. 分析:present 提出,我们在写科技论文的时候会经常用到这样的句子,可以背下来。 1.1.4


我分析的两个部分是引言和总结与写一部工作。 引言主要是将所研究的问题引出来,自己为什么要研究这个领域,这个领域前人是怎么 做的,前人的做法现在是不是遇到了什么问题,列出现如今所遇到的问题,这样才好说明白 自己为什么在这个领域上继续研究。 总结和下一步工作主要是对自己所做工作的总结, 这一部分首先要肯定自己的研究成果, 除此之外是谦虚的说一下自己所研究的还有哪一些不足之处,自己在今后的科研工作中也会 在这个方面进一步研究。

第 3 页

(全英文论文)浅析第二语言习得过程中的动机缺乏及其解决方案_教育学_高等教育_教育专区。本科生毕业设计(论文)封面( 2015 届) 论文(设计)题目 作学班论文字院、...
二语习得论文 - 第二语言习得对中国英语教学的影响 陈晓晶 【摘要】 : 本文回顾了第二语言习得理论进入中国的历史过程, 并着重阐述了它对中国英语教学的影响:1....
理解习得论文_教育学/心理学_人文社科_专业资料。理解习得论文 摘 要把听力理解...关键词 听力课堂教学;理解;习得 英语听力在英语交际中的重要性是不言而喻的。...
2014级英语专业_二语习得_课程论文选题方向和评分标准 - 此处不能书写 此处不能书写 ………装……… ...
二语习得论文 - 英语,二语习得,论文,MLA... 二语习得论文_英语学习_外语学习_教育专区。英语,二语习得,论文,MLA 成绩 本科生课程论文论文题目: Second Language ...
中间语与第二语言习得英语论文6.5_其它语言学习_外语学习_教育专区。英语专业毕业论文, 中间语,第二语言 本科毕业论文 题专班姓学 目:Interlanguage and SLA ...
二语习得论文 - 二语学习中的语言焦虑及克服 摘要:语言焦虑是影响二语学习的重要情感因素之一。语言焦虑(language anxiety)是学 习者个体与二语语境尤其是二语...
第二专业论文提纲_英语学习_外语学习_教育专区。英美留学生中文学习过程中的正负迁移现象研究摘要:自二语习得于上个世纪中期在国外发展成为一门学科以来,基于作为人 ...
英语专业毕业论文 - 江西教育学院自学考试 2010 级英语教育专业论文 交际法语言教学的必要性和重要性 自考 2010 级英语本 聂文汉 指导老师:陈忠猛 摘 要: 随着...
英语写作论文 - 成绩 论文题目 不同注释位置对词汇附带习得的影响 课程名称 授课教师 院系 学术论文写作 朱效慧 英语学院 年级专业 2014 英语(外应) 姓学时名...