Probabilistic modeling to achieve load balancing in Expert Clouds

Expert Cloud as a new class of Cloud computing systems enables its users to request the skill, knowledge and expertise of people without any information of their location by employing Internet infrastructures and Cloud computing concepts. Effective load balancing in a heterogeneous distributed environment such as Cloud is important. Since the differences in the human resource (HRs) capabilities and the variety of users' requests causes that some HRs are overloaded and some others are idle. The task allocation to the HR based on the announced requirements by the user may cause the imbalanced load distribution among HRs as well. Hence resource management and scheduling are among the important cases to achieve load balancing. Using static and dynamic algorithms, the ant colony, and the method based on searching tree all are among the methods to achieve load balancing. This paper presents a new method in order to distribute the dynamic load based on distributed queues aware of service quality in the Cloud environment. In this method, we utilize the colorful ants as a ranking for making distinction among the HRs capabilities. In this paper, we perform the mapping among the tasks and HRs using allocating a label to each HR. We model the load balancing and mapping process based on Poisson and exponential distribution. This model allows us to allocate each task to the HR which is able to execute it with maximum power using the distributed queues aware of the service quality. Simulation results show that the expert Cloud can reduce the execution and tardiness time and improve HR utilization. The cost of using resources as an effective factor in load balancing is also observed.