Adapting Market-Oriented Scheduling Policies for Cloud Computing

摘要

Provisioning extra resources is necessary when the local resources are not sufficient to meet the user requirements. Commercial Cloud providers offer the extra resources to users in an on demand manner and in exchange of a fee. Therefore, scheduling policies are required that consider resources' prices as well as user's available budget and deadline. Such scheduling policies are known as market-oriented scheduling policies. However, existing market-oriented scheduling policies cannot be applied for Cloud providers because of the difference in the way Cloud providers charge users. In this work, we propose two market-oriented scheduling policies that aim at satisfying the application deadline by extending the computational capacity of local resources via hiring resource from Cloud providers. The policies do not have any prior knowledge about the application execution time. The proposed policies are implemented in Gridbus broker as a user-level broker. Results of the experiments achieved in real environments prove the usefulness of the proposed policies.