Decision making for the selection of cloud vendor: An improved approach under group decision-making with integrated weights and objective/subjective attributes
Introduction
Cloud computing refers to an IT service model where computing services (both hardware and software) are delivered to customers on demand, over a network, in a self-service mode, independent of the device and location (Marston, Li, Bandyopadhyay, Zhang, & Ghalsasi, 2011). Compared with traditional IT, cloud computing has many advantages such as cost saving, flexibility, datasecurity, easy market entry, and faster market operations (Armbrust et al., 2010, Marston et al., 2011, Sultan, 2011).
Cloud computing has the potential to transform a large part of the IT industry, making software even more attractive as a service, and shaping the way IT hardware is designed and purchased (Armbrust et al., 2010). The use of cloud computing is now becoming more and more popular, for example, numerous government agencies and enterprises are expected to rely on the cloud for more than half of their IT services by 2020 (Garrison, Wakefield, & Kim, 2015). A recent survey by IDC suggests that public cloud computing will become a $127 billion dollar industry by 2018 (Sabi, Uzoka, Langmia, & Njeh, 2016).
Although cloud computing has been considered as a much hyped phenomenon in the IT and business world, promising to deliver a host of benefits, companies need to look beyond this hype and seriously consider the real value of incorporating cloud computing in their own businesses (Misra & Mondal, 2011). Is cloud computing really suitable for every firm? The answer is maybe not. A survey conducted by the European Network and Information Security Agency (ENISA) indicated that 68.1% of firms think that the use of cloud computing can avoid capital expenditure in hardware, software, IT support, etc., and 30.6% of firms think that cloud computing can remove economic/expertise barriers impeding the modernization of business processes by the introduction of the technology (ENISA, 2009). However some other surveys show that nearly half (48%) of the enterprises are skeptical about cloud computing, and some research even found that using cloud computing will be more expensive (Misra & Mondal, 2011). The current studies have generally acknowledged the fact that companies using cloud computing will face a lot of problems besides cost, such as technology usefulness, managerial skills, strategies, competitive pressure, etc. (Low and Chen, 2012, Martens and Teuteberg, 2012, Messerschmidt and Hinz, 2013). Therefore, firms should use scientific decision tools to judge which cloud computing vendor is more suitable by considering both cost and other related factors.
Studies in this area have just started, Misra and Mondal (2011) have done pioneering work, and they proposed a mathematical model to analyze the use of the cloud computing problem from a technology perspective and Return on Investment (ROI). Martens and Teuteberg (2012) developed a sophisticated mathematical model for the decision problem of cloud computing by considering both cost and risk factors. Low and Chen (2012) proposed critical criteria for hospitals to select a cloud based information systems provider, and used the fuzzy Delphi method (FDM) and the fuzzy analytic hierarchy process (FAHP) to evaluate the primary indicators and the weights of the criteria. Manuel (2015) introduced a novel trust model to help users to select the best cloud resources. Trust value is calculated using four parameters such as availability, reliability, turnaround efficiency, and data integrity. Püschel, Schryen, Hristova, and Neumann (2015) introduced a novel policy based on service admission control models that aimed at maximizing the revenue of cloud providers and users while taking informational uncertainty regarding resource requirements into account. Walterbusch, Martens, and Teuteberg (2015) presented a decision model for evaluating and selecting cloud computing services, which supports decision makers in comprehensively evaluating relevant cost types.
However, these studies are mainly from the technology and cost perspective, and neglect other influence factors such as competitive pressure, managerial skills and strategies, etc. Therefore, when making use of cloud computing decisions, firms should consider more attributes besides cost and technology. The attributes usually can be divided into two classes: (1) objective attributes – defined in numerical terms, e.g. cost, time, speed etc.; (2) subjective attributes – defined in qualitative statements, e.g. strategy, management, etc.
From the above analysis, we can see that the use of the cloud computing decision problem is essentially a multi-attribute group decision making problem (MAGDM). Many scholars have used such an approach to study IT/IS decision problems, e.g. Yang and Huang (2000) used the analytic hierarchy process (AHP) method to study IS decision problems, and argued that technology, economics, quality, strategy, management and quality should be considered. Lee and Kim (2000) combined the analytic network process (ANP) and AHP within a zero-one goal programming (ZOGP) model to propose an IS/IT decision method, which considers the interdependencies between evaluation criteria. Wang and Yan (2007) proposed a MAGDM approach which combines AHP and PROMETHEE to analyze the IS/IT outsourcing problem. The AHP method is used in weighting the criteria, and PROMETHEE is used for the final ranking. Chen and Wang (2009) developed a fuzzy VIKOR method to obtain the best alternative in an IT/IS selection problem. Chang, Yen, Ng, and Chang (2012) combined the Delphi and AHP methods for an evaluation model for SMEs for IT/IS decision making.
However, the above IT/IS related literature with group decision making (GDM) methods have some limitations: (1) in the decision-making process, they only consider subjective (e.g. strategy, management, etc.) or objective attributes (e.g. Investment cost), and not effectively combining them will lead to inaccuracy in the decision results, e.g. (Chang et al., 2012, Yang and Huang, 2000); (2) neglecting the attribute weights or DM weights, or assuming that they are already known. However, because every project has its own specific influence factors, and experts from different fields have their individual specific background knowledge, skills, and experience, neglect or merely subjective assignment of values to attribute weights and DM weights will increase the uncertainty of the decision-making process, e.g. (Chen and Wang, 2009, Wang and Yang, 2007).
Aimed at the above mentioned limitation (2), in the field of MAGDM research, scholars began to study how to weight DMs and attributes to make decision-making more scientific. The existing approaches can be divided into two categories: (i) subjective weighting – based on the DMs’ subjective preferences. The weights of DMs and attributes are usually given in advance, or a special evaluation matrix is established to compare the differences of the DMs and attributes (e.g. Delphi, AHP) (Cebeci, 2009, Chena and Wang, 2010, Dong et al., 2015, Efe, 2016, Zhu and Xu, 2014). The subjective weighting method requires the experts to be very familiar with each other and the decision problem, but even so, the subjectivity and uncertainty is still very strong. Therefore, academics did not undertake in-depth study in this domain; (ii) objective weighting-based on data which are given in the decision table of the attributes for each alternative. In these studies (Chin et al., 2015, Fu and Wang, 2015, Liu et al., 2015a, Wan et al., 2015, Zhou et al., 2014), the common element is that there is no need to have another evaluation matrix for the DMs and attributes, and the weights of the DMs and attributes are only computed by data which are given in the decision table for each alternative. Because these kinds of methods are more objective and accurate, it has resulted in wide research for several years.
However, as shown above, current studies about the weighting attributes and DMs in MAGDM also have some limitations: (1) both attribute weights and DMs weights are assumed to be known, or some are assumed to be known, and others are calculated. Little research has considered the situation in which attribute weights and DM weights are both unknown; (2) the weights of the attributes/DMs are computed in one of the two ways: subjective weighting or objective weighting. There has also been little research on the weighting for attributes/DMs based on a combination of subjective preferences and objective weights. However, in the actual decision situation, considering objective assessment information of decision opinions as well as the differences of the subjective preferences of DMs and their identity differences can make the decision result more accurate. Weights combining objectivity with subjectivity would be much better than those just considering subjectivity (Wang & Yang, 2007).
Therefore, in order to fill the research gap, this study proposes a cost with TOE (Technology, Organization and Environment) criteria, then a novel MAGDM approach with integrated weights and objective/subjective attributes is presented to solve the selection of cloud computing vendor problem. The proposed approach is mainly divided into the following steps: First, collect decision opinions, establish decision matrices and normalize them. Second, integrate statistical variance (SV) and simple additive weighting (SAW) to determine the attribute weights, and integrate an improved TOPSIS and Delphi–AHP to determine the DM weights; then the decision opinions provided by each DM are aggregated into the comprehensive evaluation values based on the LWAA operator. Finally, refer to the comprehensive evaluation values, and identify which cloud computing vendor is suitable for the firm to use.
The reason why we integrate these methods in our proposed approach, is mainly because of the following reasons: (1) SV, SAW and TOPSIS are more simple and superior. Zanakis, Solomon, Wishart, and Dublish (1998) compared the eight MADM methods: TOPSIS, elimination and (et) choice translating reality (ELECTRE), SAW, multiplicative exponential weighting (MEW) and four AHPs by a simulation, and found that the best performing methods were SAW and MEW, then TOPSIS and AHPs, ELECTRE were the worst. Moreover, Chang and Yeh (2001) also obtained similar results. They found the SAW and TOPSIS performance better than other methods. Chou, Chang, and Shen (2008) also stated that the simpler evaluation techniques are often superior. (2) The objective weighting method SV is more simpler than the entropy method and the Standard and Mean deviation method proposed respectively by Xia and Xu (2012) and Xu and Da (2010). The entropy method involves the computations of entropy values and finding the degree of divergence for each attribute, then dividing the divergence value of each attribute with the total value to determine the objective weights of the attributes (Rao, Patel, & Parnichkun, 2011). Therefore, determining the objective weights of the attributes using the entropy method is more complicated than the SV method. The Standard and Mean deviation method also needs more calculation steps, like standard deviation, mean deviation and solving the combined optimization model. Thus, the SV method is more simple than the Standard and Mean deviation methods, too. (3) The suitability of combining subjective and objective attributes, and using the linguistic terms to give evaluation values by DMs.
The rest of the paper is structured as follows. Section 2 proposes the evaluation criteria for the use of cloud computing decision. Section 3 introduces the proposed approach and the general steps in the decision analysis. Section 4 presents the main differences of the proposed method and the existing ones. Section 5 demonstrates a numerical example, and Section 6 presents the conclusions.
Section snippets
The evaluation criteria of the use of cloud computing
Though the number is limited, there have been some researches on the decision problem of the use of cloud computing, e.g. (Martens and Teuteberg, 2012, Misra and Mondal, 2011). However, the currently decision related studies solely focus on the quantitative factors such as cost, and neglect the qualitative factors such as strategy, management etc.
In the present study, we use the TOE framework with economic issues to analyze the performance criteria of the selection of cloud computing vendor.
The proposed method
An improved MAGDM is proposed, and the approach is mainly divided into the following three stages: (1) Preparation stage. In the preparation stage, DMs give their evaluation opinions according to each attribute by questionnaires in linguistic terms or quantitative terms. Then the decision matrix is established for each DM. (2) Aggregation stage. In the aggregation stage, a combination weighting method that considers the objective weights and subjective preferences is proposed to determine the
Comparing the proposed approach with other methods
The decision making problem in the selection of cloud computing vendor is essentially regarded as MAGDM. In the existing literature, a number of classical MAGDM techniques have been employed in this kind of problem-solving process, and can be classified them into four categories (Chai, Liu, & Ngai, 2013): (1) multi-attribute utility methods such as AHP and ANP, (2) out ranking methods such as elimination and choice expressing reality (ELECTRE) and preference ranking organization method for
Illustrative example
In this section, we use an example to illustrate the usefulness of the proposed approach. Suppose a firm wants to identify which cloud vendor is suitable for provide cloud computing technology from the comprehensive consideration of many issues. There are four cloud vendors A={A1, A2, A3, A4} to consider, and the four experts D={d1, d2, d3, d4} are respectively the CEO, an Executive in the Information Office, an Operations Executive and a Financial Office Executive. They are asked to evaluate
Conclusions
This study proposes a novel approach to solving the selection of the cloud computing vendor problem. In this approach, we first propose a cost with TOE (Technology, Organization and Environment) criteria, then a novel MAGDM approach with integrated weights (by both considering the subjective preferences and objective information) and objective/subjective attributes are presented to solve the decision problem. First, computational formulae are given to normalize the decision matrices. Second,
Acknowledgments
This research was supported by the National Natural Science Foundation of China under Project Nos. 71502159 and 71471158; The Applied Basic Research Science Foundation of Yunnan Provincial Department of Science and Technology under Project No. 2015FD028; the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. PolyU 15201414). The authors also would like to thank The Hong Kong Polytechnic University Research Committee for financial and technical support.
References (80)
- et al.
Aggregation of group fuzzy risk information in the railway risk decision making process
Safety Science
(2016) - et al.
Opportunities and risks of software-as-a-service: Findings from a survey of IT executives
Decision Support Systems
(2011) Fuzzy AHP-based decision support system for selecting ERP systems in textile industry by using balanced scorecard
Expert Systems with Applications
(2009)- et al.
Application of decision-making techniques in supplier selection: A systematic review of literature
Expert Systems with Applications
(2013) - et al.
In-depth analysis and simulation study of an innovative fuzzy approach for ranking alternatives in multiple attribute decision making problems based on TOPSIS
Applied Mathematics and Computation
(2011) - et al.
An analysis of IT/IS outsourcing provider selection for small- and medium-sized enterprises in Taiwan
Information & Management
(2012) - et al.
Evaluating airline competitiveness using multiattribute decision making
Omega
(2001) - et al.
Optimizing partners’ choice in IS/IT outsourcing projects: The strategic decision of fuzzy VIKOR
International Journal of Production Economics
(2009) - et al.
A method of determining attribute weights in evidential reasoning approach based on incompatibility among attributes
Computers & Industrial Engineering
(2015) - et al.
A fuzzy simple additive weighting system under group decision-making for facility location selection with objective/subjective attributes
European Journal of Operational Research
(2008)