A flexible deadline-driven resource provisioning and scheduling algorithm for multiple workflows with VM sharing protocol on WaaS-cloud.

Maximized deployment of workflows in the research organizations has motivated the emergence of multi-tenant environment that offer these workflows deployment as a service. Hence, Workflow-as-a-Service (WaaS) model has been made by research organizations to handle the future model of Workflow Management System (WMS) that would serve a variety of users from a specific single point of provision. These mechanisms vary from widely used WMS in that they process a variety of scientific workflows at the time of execution. A widely used WMS is mostly used to run a single workflow in a dedicated set-up meanwhile WaaS-cloud platforms improving the set-up by adopting multiple workflows deployment in a multi-tenant service model. In this study, we leverage an advanced Virtual Machine (VM) sharing protocol to optimize VM utilization and to achieve specific Quality of Service conditions from a variety of users in WaaS cloud platforms. We present a Flexible Deadline-driven resource Provisioning and scheduling algorithm for Multiple workflows (FDPM) that can minimize the computing expenses by adapting VM sharing to make the workflow's execution cost lesser while achieving a user-assigned deadline. Our analysis proves that the FDPM algorithm can leverage the VM sharing protocol to make better performance with regard to minimizing the execution cost evaluated to the modified deadline-driven scheduling algorithm. [ABSTRACT FROM AUTHOR]