Efficient job placement using two-way offloading technique over fog-cloud architectures

Fog computing contains resource-constrained devices closer to user proximity, which are insufficient for compute-intensive and large workloads. Cloud computing can help to address this limitation by transparently integrating fog and cloud data centers. However, the existing state-of-the-art does not integrate cloud and fog resources for maintaining application performance by dynamically offloading the tasks to the cloud while minimizing the cost. This paper introduces a new workload/job placement method that performs two-way offloading between cloud and fog to efficiently utilize the fog resources while minimizing the cost of using cloud resources to serve the workload. Our proposed solution is evaluated on various setups, including small-scale and large-scale data centers having homogeneous and heterogeneous physical machines (PMs) with different job batches, and compared with the existing state-of-the-art baseline method. Overall, the proposed solution outperforms the baseline method by utilizing 40.87% more fog resources while reducing × 1.68 times cloud cost.