Investigate AI-based learning for cloud services for adaptive autonomous behavior

Abstract

Cloud computing provides more reliable web services due to its flexibility for accessing resources on-demand and self-managed services. However, cloud computing faces new challenges when managing a massive amount of services in an environment full of uncertainties. Moreover, the customer´s variations of services requirements make them more complex. All of these factors increase the difficulty of managing the services on the cloud. The two keys for providing a reliable web service are service performance and resources utilization. Service performance grantees providing reliable service in terms of response time, but that should be achieved without wasting the limited amount of available resources. Therefore, providing an optimized service without under- or over-provisioning is a requirement. There is a need to develop an autonomous service to adapt itself based on the surrounding environment. This thesis explores introducing a learning automaton algorithm for developing an autonomous web service. The result is aSpace machine that has been built and developed through a set of phases and shows a progression from an autonomous scalable service towards the aSapce machine. The aSapce machine optimizes the service performance in terms of self-provision. The results indicate the ability of the aSapce machine to manage the different workloads effectively by providing an elastic web service. Further exploration for this can pave the way towards handling more complex situations.