Funding of cash market cloud resources with a cost-based method

Dissertation for M.Sc degree

Computer Engineering - Software

Abstract:

The rapid growth in demand for computing power has led computing to move towards a cloud computing model based on massive virtualized data centers. Cloud computing allows users to provide computing resources efficiently and dynamically to meet their needs. The use of these resources certainly requires payment by the users. Most cloud service providers offer two payment methods, reservation and on-demand. The reservation method is cheaper, but due to the uncertainty of the user's needs, this method cannot be used only, and sometimes the amount of reserved resources may be less or more than the user's actual request. The more accurate prediction of the future needs of users and, as a result, the more accurate the amount of resource reservation, leads to a reduction in the overall cost of providing resources in the cloud. In this thesis, multilayer perceptron neural network with error back propagation training model is used to predict future needs more accurately. Then, the colonial competition algorithm has been used to allocate resources to suppliers in order to optimize the final cost. The results obtained from the simulator designed in the NetBeans environment show a reduction in the cost of providing resources compared to previous algorithms.

Introduction:

Before the advent of computers, if a person wanted to find information, he had to physically refer to certain sources and spend a lot of time and money to find and use them. With the spread of computers and digitization of information, the process of finding and using information became easier. But with the advent of the Internet and web-based services, a very valuable evolution was created in the field of information technology. Internet, which can be used to send names to the farthest part of the globe in a few seconds, find the list of images and information required in a few thousandths of a second, manage and monitor your business from anywhere, documents and information are always available and connected with others at all times.

Another demand that has always been discussed by information technology researchers was the long-standing dream of being useful as a service[1], which was realized with the advent of virtualization technology[2]. The possibility of using computing resources in the form of a virtual machine [3] has been provided in this technology. Integration of servers and optimization of infrastructure, better and easier management and security of servers, improvement of resource efficiency, reduction of costs and easy upgrading and updating are among the important advantages of virtualization technology [1]. Today, computing is provided as a service in the form of cloud computing [4] by using virtual machines [2]. Technologies such as clustering [5], mesh [6] and now cloud computing are all moving towards achieving very high computing power in a completely virtual way by pooling resources and providing a single and integrated system in the eyes of the user. In addition, the important goal of these technologies is to deliver computing on demand. Beneficial computing [7] describes a model of business and commerce in which computing power is provided on demand. Customers pay an amount based on the amount of use of the service to the provider, similar to the way we see today in public utilities such as water, electricity, gas, and telephone, where the amount of payment is determined according to the consumption of each of them in the respective bills [1].

It can be clearly seen that after nearly two decades of the emergence of the Internet and years of hardware, software, and service supply, a fundamental question has suddenly formed in the minds of researchers and thinkers in this field.

Cloud computing services are based on the "pay-per-use[8]" model, and large companies provide these services [3], [4], [5], [6].

But what is cloud computing? What are the differences with previous computing and what challenges do cloud providers and consumers face? How can the cost of using resources be optimized for users as much as possible?

In the first chapter of this treatise, the general concepts related to cloud computing are introduced. In the second chapter, the problem, history and previous works in the field of reducing the cost of providing cloud resources have been discussed. The third chapter describes multi-layer perceptron algorithms with error backpropagation training method and colonial competition. In the fourth chapter, the proposed method has been presented and evaluated using a simulator, and in the fifth chapter, the conclusion has been discussed. 

Chapter 1

A review of cloud computing and its related concepts

1-1) The concept of cloud computing

The emergence of the basic concepts of cloud computing goes back to the 1960s [7]. When John McCarthy stated that "computing may one day be organized as one of the public industries".

Almost all the features of cloud computing today, along with a comparison with the electricity industry and forms of public, private, government and association consumption, were examined by D. Parkhill in a book entitled "The Problem of Public Computer Industry" in 1966 [8]. The word cloud is actually derived from the telephone industry, as the telecommunication organizations, which until the 1990s only offered dedicated point-to-point lines, started offering virtual private networks with similar quality and lower prices. The cloud symbol was used to show the border point between the parts that are the responsibility of the user and those that are the responsibility of the supplier. Cloud computing expands the concept of the cloud to include servers in addition to network infrastructure. [9]

Amazon played an important role in the expansion of cloud computing by modernizing its data center. They found that by changing their data center, which, like most computer networks, used only 10% of its capacity most of the time, and the rest of the capacity was reserved for short periods of peak use, they could improve their internal efficiency with cloud architecture. Since 2006, Amazon has provided access to its system through web services based on public computing. In 2007, Google and IBM together with several universities started a large-scale research project in the field of cloud computing. In the middle of 2008, Gartner realized that there is a situation in cloud computing that arises to "shape the relationship between the consumers of information technology services, between those who consume these services and those who sell these services" [10]. So, in order to reach a new beginning in this market and to replace web-based programs instead of software related to host servers, this term has been created. Describing why the word "cloud" is used in this term is also very simple. In computer network diagrams, the Internet is usually shown as a cloud in the image. The reason for this analogy is that the Internet, like a cloud, keeps its technical details hidden from users. So, in fact, this phrase has provided a way for consultants and companies to sell their services in new packages. Since commercial companies are transferring their software on the web and these programs are exposed day by day with new and more interesting features through browsers, it can be said: soon we will be able to access everything from any browser and with any computer without any boundaries between the personal computer and the Internet. Now let's define cloud computing in detail and define its key features. 1-1-1) Definition of cloud computing Many academic and commercial researchers try to define exactly what cloud computing is and what unique features it offers.