A decision-based pre-emptive fair scheduling strategy to process cloud computing work-flows for sustainable enterprise management

Abstract

In cloud computing, clients comply a policy of pay-as-you go, i.e., they only pay for the resources they use. So, the processing power of the clouds has to be optimised to reduce the cost at client's side. Using the resources optimally ensures enterprise sustainability of cloud service providers. Workflow is a set of tasks that are interdependent on each other. Scheduling these workflows is one of the most important challenges to optimally utilise the cloud resources and ensure better quality of service (QoS) to clients. Existing works on scheduling in cloud computing mainly focus on scheduling independent tasks rather than (inter)dependent tasks. In this paper, we propose a strategy to schedule dependent tasks called pre-emptive fair scheduling algorithm (PFSA). This is fair scheduling strategy also aims to ensure higher utilisation of virtual machines (VMs) by reducing the idle time and to minimise the number of times a pre-empted task is submitted to the virtual machine. In both cases, this algorithm tries to effectively reduce the overall processing time of dependent tasks at virtual machine, thus minimising the cost involved in processing of tasks. This economically viable decision-based strategy will be helpful for cloud service providers in ensuring sustainability.