Effective Planning Algorithm for Meeting Parallel Scientific Application Deadline Requirements in Edge-Cloud Platform

In recent years, complex scientific workflow's have gained more attention. These scientific workflow's usually comprise directed acyclic graphs that have interdependency among different tasks. For the faster computation of these scientific workflow's, they are executed in heterogeneous clouds, which provide more resources for execution and do not focus on reducing energy and makespan. To address the problem, i.e., reducing energy and makespan, edge-clouds have emerged. The edge-clouds reduce the cost, energy, and makespan, yet fail to provide reliability because the edge-clouds use different physical machines and virtual machines to provide resources. To address this problem, Virtual Network Functions (VNFs) are being utilized, which provide better reliability but fail to reduce energy usage and makespan. Hence, this work presents an Effective Planning Aware-Scheduler (EPAS) for reducing energy usage and makespan, assuring optimal service provisioning by sorting tasks based on their completion makespan and scheduling tasks using multi-objective optimization. Further, an energy minimization offloading mechanism is presented to offload tasks between cloud and edge-server. This work has been evaluated using SIPHT scientific workflow's where the makespan, energy, and power have been reduced by 39.32%, 69.14%, and 56.57%, respectively.