N = (n − 1)/2,
where N is the number of pairwise comparisons, and n is the decision elements at every level.
3.2.Inconsistency
Inconsistency occurs when there is an intentional or unintentional error while performing a pairwise comparison by an expert. There are two types of inconsistency: ordinal and cardinal [6]. In ordinal inconsistency, the ranking order of elements should be upheld. For example, if someone likes apples more than oranges, and oranges more than grapes, then that person should like apples more than grapes. However, if that person prefers grapes over apples, then that is accounted as ordinal inconsistency. Cardinal inconsistency occurs when the element’s proportion is not upheld. For example, if someone regards apples as being two times more valuable than oranges and oranges as being three times more valuable than grapes, then that person should regard apples as being six times more valuable than grapes, or else cardinal inconsistency could be observed. It is observed that an inconsistency of 10% (or 0.10) is considered as an acceptable inconsistency [3].
3.3.Disagreement
Unlike inconsistency, disagreement is calculated based on the differences between the opinions or evaluations of the expert panel. If the disagreement among the expert panel is beyond a certain range (which is considered to be 10%), then a brief session should be conducted with the experts to try to convince that particular expert whose judgment was different from the rest. There are additional tests such as an F-test that can be performed to determine whether the disagreement among experts in a panel is statistically significant or not [6]. “Understanding and resolving the disagreement is an important aspect of the research and for building the decision model” [6].
3.4.Decision model
The objective of the model in this paper is to choose the best cloud computing service provider for an application developer.
Figure 1 illustrates the process of decision-making for this paper using a HDM.
Figure 1.A HDM’s process of decision-making.
3.5.Expert panel
An expert panel consists of application developers (across different industries) who use different cloud computing platforms. All the experts are significantly related to the model’s objective and decision elements. The experts are from different backgrounds, age and sex.
3.6.Decision elements and model levels
Each of the 13 experts accessed the HDM software and gave their professional judgments and quantification with regard to each of the following criteria:
1.Innovation: Is it easy to scale up the cloud if the need increases or is it compatible with the other applications?
2.Technological: What is the architecture of the platform or how secure is the cloud?
3.Usability: What is the platform’s ease of use or is the chosen platform useful for a certain application?
4.Economical: What is the total monetary expense of the proposed cloud computing platform?
•Level 1: Mission (selecting a cloud computing platform for an application).
•Level 2: Four Primary Evaluation Criteria using a pairwise comparison method. A total of a hundred points was divided between the two criteria in proportion to their relative importance to the problem objective.
•Level 3: Seven Secondary Evaluation Criteria with respect to the Level 1 criteria using a pairwise comparison method. A total of a hundred points was divided between the two criteria in proportion to their relative importance to the problem objective.
•Level 4: Four cloud computing service providers using a pairwise comparison method. A total of a hundred points was divided between the two criteria in proportion to their relative importance to the problem objective.
3.7.Cloud computing complexity
A cloud system is used to sustain effective teleworking and tools establishment supporting virtual teams of employees situated around the globe. Huczynski and Buchanan [7] define a virtual team as a team “that relies on technology mediated communication, while crossing boundaries of geography, time, culture and organization to accomplish an interdependent task” [7]. It can be said that every team functions in its own way. Additionally, physical separation of the team members across the boundaries will be different for each team, not only making it more unique than other teams but also creating inconsistencies within as teams grow and evolve over time. Shin [8] argues that there is a range or scope of “virtualness”— the larger the dispersion in a team, the more amplified is this virtualness [8]. Hence, it is very important that there should be an effective way of interaction, communication and level of engagement between team members, in order to implement successful virtual teams. This is the most vital yet challenging part of team management for such teams. According to Duart and Snyder [9], critical success factors in managing virtual teams and facilitating effective interaction between members consist of the right selection of IT tools, team members’ competencies, the leadership and culture of the team, the process standardization and training of team members towards these goals [9].
3.8.Cloud computing compatibility
The existing competition between the big service providers of cloud computing has made the service incompatible. Current provided solutions by these vendors are not really compatible with each other [10], as they tend to lock in existing customers into their provided services or infrastructure, and prevent data or software portability [11]. In addition, dominant vendors are not willing to accept the common standards, which ultimately results in incompatible platforms [12], which again increases the lock-in effect. This greatly prevents many small and medium businesses from entering the cloud market. The European Network and Information Security Agency (ENISA) and European Commission states that the vendor lock-in problem is a high risk that cloud infrastructures are facing [13]. It can be said that cloud compatibility is the solution for this problem, which will not only improve this situation but also benefit both consumers and providers.
Cloud