Agent Enabled Adaptive Management of Cloud Service Provisioning

Infrastructure as a Service (IaaS) providers tend to use heterogeneous hardware environments and virtualization techniques to provision a wide variety of virtual machines that have varying specifications, exhibit varying performance levels, and are offered under different pricing models. Consumers also have very diverse and dynamic infrastructure requirements depending upon the workloads that they intend to deploy on the cloud. Given the diverse and dynamic nature of demand-side requirements and supply-side capabilities, market participants can benefit from decision-support and decision-making tools that automate, augment and coordinate some or all of their decision processes leading to the service exchange. This can be realized through autonomous software agents that can offer the necessary capabilities for adaptive management of the different tasks associated with service provisioning. We use the practical example of on-demand resource provisioning on Amazon EC2 to motivate, illustrate and validate our proposed solution.