标题: 运用云端运算在系统容易出现迫切需求: 用亚马逊Web服务(AWS)
Applying Cloud Computing to systems prone to pressing demand: Using Amazon Web Services (AWS)
作者: 黎高昆
Le, Cao-Con
林宝树
Lin, Bao-Shuh Paul
资讯科学与工程研究所
关键字: 云端运算;选课系统;火车票系统;亚马逊Web服务;Cloud Computing;Course Selection System;Railway Ticket System;Amazon Web Services;AWS
公开日期: 2013
摘要: 云端运算的使用,在技术及商业界已经司空见惯。云端服务供应商的数量一直不断增长,因此有一系列给远端用户的选择。也有一些优秀的应用,如Dropbox的发明,档案托管服务,提供云端储存以及文件同步等等。
藉由一些课程,我学习到与云端运算相关的大量的知识,我参加了一些课程并做一个敏锐的观察者,我意识到,这样的系统如选课系统(例如,在交大的选课系统)或在我的国家,越南的铁路销售系统,像这样的系统有时不太稳定。
当网站流量激增、流量过量时,将造成伺服器负载急剧增加而影响伺服器上的网路连接。比方说针对选课系统来说在学校让学生选课时有时候会有超多学生同时一起来跟系统连结,这会造成伺服器当机或是反应很慢。另外一个系统像铁路销售系统在过年时有许多人为了工作而离开家里从北部移到南部,然后过年时要回家团圆,因为他们同时跟系统连结,所以这也会造成一个系统像上面提到一样的。
针对类似上面提到的系统我们会有什么办法又简单又便宜能够解决短时间迫切需求之系统呢?
亚马逊Web服务(AWS)是一家云端运算供应商,他们提供许多服务又便宜又有大弹性。所以我认为这就是我们的选择之一。在我的研究,我建好一个简单的系统跟选课系统差不多,然后把这个系统移转到远端去,接下来是针对这两个系统做评估看它们的效率如何。最后我建议在云端可以建一个系统,它的架构包含一个域名系统Web服务、三个负载平衡器、三到十八台虚拟Web伺服器、和一台资料库伺服器。Web伺服器的数量会变动是因为靠使用者的数量同时供用让整个系统做自动缩放。假设我们本来的系统频宽大概1Gbps,然后跟新的系统做一个比较,新的系统能力最大可以比本来的系统能力高三倍。而且新的系统也不算贵,每学期大概要花美金600元。
完成这个研究,我有一个结论是上面类似之系统可以改善它的情况如果有应用云端运算。
Cloud computing has been a commonplace and widely used in both technological and business world. The number of cloud providers has been constantly growing and so has the set of options for end users. There have also been some outstanding applications such as the invention of Dropbox, a file hosting service that offers cloud storage and file synchronization.
With a great deal of knowledge on cloud computing from some courses that I took in the master program and as a keen observer, I realized that such systems as the course selection system (for example the one that operates at NCTU) or the railway ticket system back in my country, Vietnam, do not work well sometimes.
There are times that websites are overloaded to a point when their services are degraded or disrupted entirely. This web traffic overload happens when there is a large surge in traffic to a particular website causing a dramatic increase in server load and putting severe strain on the network links leading to the server. For example many students access the course selection system at the time the school lets the students select their courses on it. Also, many people want to buy railway tickets on railway ticket system at the time closing to New Year vacation when they want to go home.
For those systems, is there any practical way to improve them with low cost but high degree of effectiveness?
Amazon Web Services (AWS), a cloud computing provider, provides a number of services with low cost and rapid elasticity. Therefore, this should be considered as a choice for us. In this research, I am going to conduct several experiments with an aim to migrate a sample system like a course selection system to AWS and evaluate the effects of the new system. Furthermore, I expect to propose a model for the new system on the cloud which includes one domain name system web services, three load balancers, three to eighteen web servers, and one database server. The number of web servers might be changed depending on how many users access the new system at the same time. It’s assumed that the original system has the capacity bandwidth of 1Gbps. As well as that if we compare the new system with the original one, the capacity of the new system is approximately 3 times as big as that of the original system. Furthermore, the cost of new system is relatively low estimated at $600 for every semester.
After doing this research, I have reached the conclusion that the overload system may be improved by applying cloud computing technology.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT070056137
http://hdl.handle.net/11536/73000
显示于类别:Thesis


文件中的档案:

  1. 613701.pdf
  2. 613701.pdf

If it is a zip file, please download the file and unzip it, then open index.html in a browser to view the full text content.