Trust-A ware Resource Provisioning for Meteorological Workflow in Cloud

摘要

Cloud computing centers are becoming the predominant platform of offering high-performance computing services based on high-performance computers. However, enabling meteorological workflow that requires real-time response is still challenging due to uncertainty in the cloud, once the computing nodes in the cloud are down, the tasks deployed on the cloud will not be completed in time. To address this problem, an optimal cloud resource for the downtime tasks provisioning method (ODPM) is proposed by formulating a programming model. The ODPM method can select the appropriate migration strategy for the tasks on the compute node to achieve the shortest workflow completion time and load balancing of the compute center compute nodes. A large number of experimental are conducted to verify the benefits brought by ODPM.