Auction Based Resource Provisioning in Cloud Computing

Cloud computing has evolved through various phases from distributed computing, mobile computing, grid computing and currently is in most evolved form of the computing. Cloud computing is basically a business model where various hardware resources such as CPU, network, storage and software resources such as platform, applications are available on demand and are offered as utility; i.e., a Cloud user has to pay for the usage of the Cloud resources. The proliferation of the Internet has made possible the availability of the Cloud services across the globe, though private Cloud exists on intranet. As Cloud is a business model, a large number of players have joined the Cloud platform with their services. Recently, a tough competition has been noticed in the Cloud market for the offered services as both the number of providers and customers have increased to a large extent. Though it creates a good opportunity for both provider and customer, it also results in a number of issues to be addressed properly to make the Cloud market more beneficial for the customer and the provider. This chapter takes a look at widely accepted Cloud computing definitions, explains its concepts and summarizes various services and deployment models of Cloud computing. This chapter also gives an overview of benefits and challenges of the Cloud computing. Further, it describes pricing in detail to understand the need for auction of the Cloud resources followed by a mention on the organization of the book. The aim of this chapter is to offer a prerequisite to understand the concepts discussed further in the book and its role in research.

Auctioning has a long history and is reported to have been used in Babylon as early as 500 B.C. The entire Roman Empire was sold off using auction in 193 A.D. [83]. With time, auction theory has evolved with more sophisticated and mature auction procedures. Auction is considered as an efficient and fair mechanism as it provides equal opportunity to both the seller and the buyer. The price of the resources is decided on the basis of the value of the resources for the bidders that makes higher revenue. In early auction days and even during its evolution, only antiques and art matters were sold using auction but now various commodities, e.g. fish, bond, spectrum, computing resources, are sold using auction because of its multifaceted benefits. After the introduction of mechanism design, auction has become a great success in economics for the resource allocation. Mechanism design and information and communication technologies (ICTs) are the two major responsible factors that made game theory and optimization an effective tool for auction design to achieve specific goals. In economics, the rich literature and practical implementation on auction are available. Because of involvement of pricing in resource allocation in Cloud computing, good possibilities can be explored for applying auction in Cloud computing. Academicians have proposed various types of auction models that can be applied in different scenarios in Cloud computing. Spot market has been become a milestone for both, academicians and professionals, to explore auction in greater depth. Amazon, a giant in the Cloud computing market, practically gave a push to implement dynamic pricing using auction. Plenty of works on auction theory, variants of auction as well as its applications for suitable scenario are available in the form of books, survey papers, etc. The aim of this chapter is to provide a detailed description of most important findings of the auction so that academicians and researchers can work their way through these findings in Cloud computing.

Forward auction, also known as “sell-side auction”, has multiple buyers and single seller in the auction market. The buyers compete by bidding their valuations for the required quantity of resources. In Cloud computing, Amazon sells its unutilized capacity using forward auction. In this chapter, a description of combinatorial forward auction with their relevance and applicability in Cloud is discussed in detail. A combinatorial auction-based formulation is given for the VM provisioning and pricing problem in Cloud market. Some possible allocation algorithms and pricing schemes are discussed for the formulated problem. A detailed survey of existing combinatorial auction mechanisms in Cloud computing is given. Finally, research challenges and issues are discussed to explore future possibilities.

Out of many auction techniques, reverse auction is the one in which providers act as bidders. In this chapter, a detailed description of reverse auction and its benefits to Cloud providers and customers are discussed. Various attributes of reverse auction are presented in the Cloud scenario for its applicability. Various auction properties, an auction designer vies to achieve, have also been listed and deliberated. A detailed and comparative survey of existing works in reverse auction on the basis of auction properties is given to better understand the state of the art of reverse auction in Cloud. A truthful combinatorial reverse auction model is also presented to help in understanding and designing a reverse auction mechanism for the Cloud researchers in order to achieve a defined goal. Open research issues are highlighted to provide an insight for the future research work in this area.

In double auction, both the user and the provider bid in the market. In this chapter, a detailed description of double auction mechanisms and their applicability in Cloud market is discussed. Various auction properties, that a double auction mechanism should satisfy, have also been discussed and deliberated. A detailed survey of related work in double auction-based Cloud resource provisioning and pricing is provided. A truthful multi-unit double auction-based model is also presented to help and motivate the Cloud researchers/academicians to design such efficient and truthful mechanisms for the Cloud market. Finally, research challenges and issues are identified with future research possibilities.

Amazon provides virtual machines using auction called spot instances besides providing virtual machines with on-demand and reserved policy. Spot instances are real implementation of auction in Cloud computing. In this chapter, first a brief description of Amazon spot market is given to make readers understand that whatever they learnt from this book are going to fetch enormous feasible opportunities in research. The chapter briefly recalls key concepts and key learning points, obtained from this book, along with summarization. Auction itself has a rich literature, and it has not been applied only in economics but in areas in Computer Science also. This chapter brings forth various literature related to auction and its application in other fields of Computer Science for further research.