A Prediction-Based ACO Algorithm to Dynamic Tasks Scheduling in Cloud Environment

In recent years, cloud computing has gained more and more attention. Task and resource scheduling becomes one of the key problems in cloud computing. This paper refers to a new prediction-based algorithm based on Ant Colony Optimization which combines the available computing resources (referred as Virtual Machines, VMs) with arriving jobs with various Quality of Service constraints (QoS, defined by users). The traditional Ant Colony Optimization algorithms usually contain properties of computing resources but without taking users' constraints into consideration and ignore the heterogeneity of cloud resources. Therefore, an algorithm in this paper is proposed which classifies the jobs into two species. And then users' QoS constraints are sorted as well as computing resources according to their computing capabilities. This paper aims at proposing an ant colony optimization (ACO) algorithm to schedule jobs with various QoS parameters on VMs with different resource parameters. Experiment results show that the proposed prediction-based algorithm outperforms the ACO algorithm to some extent in finding the best dispatch of tasks to VMs.