Improving coordination of decentralized self-adaptive system in Multi Cloud environments

Managing coordination between Autonomic Managers in the Multi-Cloud environment requires an effective control mechanism to handle issues such as conflict during decision-making and in the execution phase. MAPE-K control loop provides a primary control loop in managing adaptation activity where the complex environment especially in Multi-Cloud with decentralized Autonomic Managers, is not sufficient to handle multiple adaptation decision with necessary control. Applying negotiation and synchronization in planning and execution phase help to resolve decision conflict issues between decentralized managers. Negotiation allows two parties to discuss in order to meet particular agreement while synchronization controls execution process according to agreed negotiation result. This paper introduces a new coordination model by combining negotiation and synchronization during planning and execution phase accordingly based on the MAPE-K control loop. This new coordination model is evaluated by using enhanced CloudSim simulator, which covers a decentralized self-adaptive Multi-Cloud environment. Result from the experiment shows that combination of negotiation and synchronization are capable of managing conflicts during a runtime adaptation.