A method to arrive at configuration consensus in a multitenant shared-service cloud environment is presented in this paper. Section 1 discusses the multitenant cloud. The authors are interested in such systems where each member of the user community has the capability to dynamically modify or preselect various preferences. Since an activity on the part of one user might conflict with the preferences of another member of the community, the authors propose user-centric adaptations based on preference-based analyses. They introduce a four-fold process: monitoring, analysis, planning, and executing (MAPE) in terms of a loop. They use a game-theoretic mechanism to demonstrate their approach.

Section 2 discusses the motivating scenario. Section 3 covers the problem. Section 4 describes a solution approach. Section 5 presents the implementation of their prototype.

While this paper could be of value to workers in the field, several comments are in order. The scenario chosen as an example comprises four tenants and their preferences. The preferences are a combination of the mundane and the extraordinary. The choice between “Aero” and “Classic” is mundane. The choices among “Very frequent antivirus checks,” “Frequent antivirus checks,” and “Unfrequent virus checks” are strange and seemingly forced, as is the choice between “Highest firewall level” and “Medium firewall level.” These choices attempt to reflect real-life conditions, but they seem artificial. Indeed, a reader could code the preferences (and the authors do in some cases) differently; for example, “Aero” and “Classic” could be coded as “x” and “not x” while the preferences “Very frequent antivirus checks,” “Frequent antivirus checks,” and “Unfrequent virus checks” could be coded as “y+,” “y,” and “-y.” The analysis could proceed without the semantics.

Three very complex figures are included purporting to indicate how the process works. Section 4.2, “Analysis,” uses game theory to arrive at a solution. Unfortunately, the discussion is chock-a-block with formulae and must be either taken at face value for those who are not game theory adepts or taken for granted by adepts.

The results of the game are discussed as it progresses. The authors discuss the need to extend their work and give examples of where they may choose to go.

In the final section, as they point out, their approach could equally apply to “any kind of adaptive system” and they chose one based upon multitenant services. This work could be of general utility; perhaps they chose the multitenant environment to attract attention to the effort. A more abstract presentation would have been valid, but perhaps less noteworthy.