TRANSACTIONS ON SERVICE COMPUTING, VOL. *, …

1 1939-1374 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See for more article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI , IEEET ransactions on Services ComputingTRANSACTIONS ON SERVICE COMPUTING, VOL. *, NO. *, * 20151A Fund-Constrained Investment Scheme forProfit Maximization in Cloud ComputingKenli Li,Member, IEEE, Jing Mei, and Keqin Li,Fellow, IEEEA bstract Cloud computing is becoming more and more popular and has received considerable attention recently.

2 As a new kind ofInformation Technology (IT) commercial model, understanding the economics of cloud computing becomes critically important. Fromthe cloud SERVICE providers perspective, profit maximization is the top issue for them. Because a multi-server system is devoted toserving one type of SERVICE requests and application, SERVICE providers should build multiple multi-server systems to satisfy the marketrequirements of different application domains. Because available funding for a SERVICE provider is generally limited, it cannot afford toinvest in all application domains.

3 Hence, how to select appropriate application domains for investment and allocate funding such thatthe total profit is maximized are important issues for SERVICE providers. To address this problem, a fund-constrained profit maximizationmodel is proposed. However, the exact solution of this optimization model is very difficult to formulate due to its complexity. Hence, thispaper presents a heuristic strategy to search for a high quality solution. In our strategy, the optimization problem is solved in fourstages, and the solution is optimized gradually.

4 Through the proposed heuristic investment strategy, an appropriate investment schemecan be developed that synthesizes the market requirement, the fund constraint, the SERVICE level agreement, and so forth. A series ofnumerical calculations is executed to assess the performance of the proposed strategy. Then, six other investment strategies arecompared to our strategy. Our results show that the investment scheme designed using our strategy can produce much more profitthan these six other Terms Cloud computing, deadline, fund allocation, multi-server system, profit maximization, queuing model, SERVICE -levelagreement, waiting computing is becoming more and more popu-lar and has received considerable attention new kind of IT commercial model, cloud computingcentrally manages the resources and offers resources andservices over the Internet to customers on demand [1].

5 Thiscan reduce the requirement for large capital outlays forthe hardware to deploy SERVICE and the human expenseto operate it [2]. Of course, the services provisioned bycloud computing are not free. The cloud SERVICE providerssupport the operation of cloud computing and earn profitsby charging the customers who enjoy the services accordingto various pricing models [3, 4, 5].In a cloud computing environment, there are alwaysthree tiers of participants: the infrastructure providers, theservices providers, and the customers (see Fig. 1 and it-s further elaboration in Section ).

6 The infrastructureproviders maintain the basic hardware and software facili-ties. The SERVICE providers rent resources from infrastructureproviders and provide services to customers. The customerssubmit their SERVICE requests to the SERVICE providers andare charged based on the quantity and the quality of theprovided services [6]. In this paper, we focus on the research JingMei is with the College of Mathematics and Computer Science,Hunan Normal University in Changsha, Hunan, China, : Ken li Li and Keqin Li are with the College of Information Science andEngineering, Hunan University, and National Supercomputing Center inChangsha, Hunan, China, : Keqin Li is also with the Department of Computer Science, State Univer-sity of New York, New Paltz, New York 12561, received **, 2015; revised **, SERVICE providers.

7 For most businesses, profit is the mostimportant concern. Hence, how to maximize profit is thetop issue for SERVICE providers. Profit maximization is anoptimization problem that is a classical issue in clustercomputing and grid computing [7, 8, 9, 10, 11]. With thedevelopment of cloud computing, understanding the eco-nomics of cloud computing becomes critically [12, 13], cloud computing economics is introduced as asignificant, new research topic in computer are many factors that affect the profit of the serviceproviders such as market demand, capital input, configu-ration of the cloud SERVICE platform, pricing model, andso forth.

8 For an application domain with a known marketdemand, the cloud SERVICE system must be configured ap-propriately to achieve profit maximization. If the capacity ofthe SERVICE system is configured to satisfy the peak demand,over-provisioning can occur because the available resourcesmay exceed the actual demand most of the time; hence, asignificant quantity of resources would be wasted. If theservice system is configured with a lower capacity, under-provisioning can occur as the available resources are unableto fully meet the fluctuating demand, which also leadsto profit loss.

9 It is important for the SERVICE providers totrade off between the two extremes to maximize the achieve this goal, Caoet al.[1] proposed an optimalconfiguration strategy for the SERVICE system. However, theauthors focused only on the profit maximization problem ofa single domain and did not consider the funding this paper, we study a different issue from that dis-cussed in the existing researches. In practice, many serviceproviders want to configure their own cloud serving plat-forms by renting resources from infrastructure providers,but their available funding is limited.

10 There are so many ser-1939-1374 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See for more article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI , IEEET ransactions on Services ComputingTRANSACTIONS ON SERVICE COMPUTING, VOL. *, NO. *, * 20152vice domains from the end-users that the available fundingis not sufficient to support all serving domains. Hence, howto utilize the limited funding to obtain a maximum profitis an important and interesting problem.