Scheduling Linear Workflows with Various Levels of Privacy Requirements in a Fog – Cloud Environment
Helen D. Karatza · 2023
In complex distributed computing architectures, such as fog and cloud frameworks, there are workloads which consist of jobs characterized by different levels of privacy requirements. In this type of computing environment, it is of paramount importance to employ privacy-aware scheduling schemes, so that appropriate execution of the applications is ensured, and the desired Quality of Service (QoS) is achieved. Based on this direction in this research we study privacy-aware scheduling techniques for linear workflow (LW) applications in a fog-cloud computing platform. Only the fog resources are considered suitable for processing LW jobs with privacy requirements. The remaining LW jobs may run on either cloud or fog resources. A new scheduling technique is employed which applies approximate computations for the sake of jobs requiring to ensure privacy of their data. It is compared with the case where no approximate computations can be tolerated. Extensive simulation experiments are carried out for the evaluation of their performance. The simulations reveal useful insights into how different levels of approximate computations set dynamically at run time affect the performance of the system in each one of the investigated cases.