Cost adaptive workflow scheduling in cloud computing

In cloud computing, it remains a challenge to allocate virtualized resource with financial cost minimization and acceptable Quality of Service assurance. In general, the VM instance is allocated to cloud service users based on not actual job processing time but the fixed resource allocation time predetermined by cloud pricing policy in contrast to grid environment. In this case, the unnecessary cost dissipation is occurred by the wasted partial instance hours of allocated resource. To address this problem, we propose the heuristic based workflow scheduling scheme considering cloud-pricing model in this paper. Our scheme is composed of two phases: VM packing and MRSR (Multi Requests to Single Resource) phases. In VM-packing phase, preassigned multi tasks are aggregated into the common VM instance sequentially, and these tasks are merged in parallel by MRSR phase. By using our proposed schemes, we are able to reduce the number of required VM instances and achieve the significant cost saving while we guarantee the user's SLA (Service Level Agreement) in terms of workflow deadline. Our proposed schemes cannot only reduce the cost by 30% compared to traditional workflow scheduling schemes but also assure user's SLA.