Orchestration of Real-Time Workflows with Varying Input Data Locality in a Heterogeneous Fog Environment

As fog computing continues to gain momentum, the effective orchestration of Internet of Things (IoT) workloads on such heterogeneous environments is an open challenge. In order to devise effective load balancing and scheduling strategies, it is crucial to gain a deeper understanding of how the locality of the workload initial input data affects the performance of such platforms. To this direction, in this paper we evaluate the performance of a heterogeneous fog environment where multiple data-intensive workflow jobs with deadline constraints arrive dynamically. Each entry component task of a workflow may require input data either from the IoT layer or from local fog resources (e.g., IoT data that have already been transferred to the fog layer or data processed by previous jobs). We investigate the impact of workflow entry task input data locality on the performance of the system, under different data location probabilities. This is the main goal and contribution of this work. The evaluation of the framework under study is performed via simulation, in an attempt to gain useful insights into how the locality of the workload input data affects the adopted performance metrics.