A Balanced Approach to Resource and Revenue Optimization in Cloud Computing using Stackelberg Equilibrium and Technical Analysis

Abstract

Cloud computing is a ubiquitous solution to several problems in the business information technology domain. It provides ready infrastructure, resources, and solutions to cater needs of thousands of organizations. New-age businesses can save millions by using subscription-based cloud service model instead investing in large capital costs in IT infrastructure. Cloud computing is a developing concept and provides opportunities to investigate and to devise new models, frameworks, and ideas that may improve the cloud ecosystem that benefits all the stakeholders. newlineCloud computing is a multi-tenant environment; multi-tenancy introduces the problem of resource allocation, i.e. sharing resources across many users, hence resource allocation is the most important problem in cloud computing environment. The most common goals of resource allocation problems are the availability of resources and revenue generated from resources according to various QoS requirements as described in SLA. Resource management is more holistic approach, and it involves more factors. Resource management proves more efficient than resource allocation in longer term. newlineResource allocation problem can be redefined as a general assignment problem. It is a form of a common problem which occurs repeatedly in day-to-day transactions in many businesses; in various forms including reservation problems, transportation problems, product mix, assignment problems, etc. A mechanism based on a rule-based constraints model can solve this problem efficiently. newlineThis problem once formulated as a pattern can help solve many problems with similar types of constraints. A strategy class can control behaviour of this pattern, strategy class should dynamically change the nature of allocation based on heuristic data. This rule-based pattern can implement rules including various QoS parameters like availability of resources, response time (including the makespan, turnaround time, waiting time, completion time, burst time etc.) throughput, costs (including energy cost, operationa