Cost-Aware Computation Offloading for Managing Cloud-Bursts in IoT-Based Cloud-Fog Networks
Sujit Bebortta, Om Prakash Pal, Dilip Senapati, Subhranshu Sekhar Tripathy · 2024
The concept of cloud bursting is a method used in mixed-hybrid cloud computing environments in the context of service outages. When a massive amount of workload hits the computing servers at once, it leverages public resources through outsourcing to assist private cloud users. The idea of fog computing can be easily leveraged to include processing several small workloads from Internet of Things (IoT)-connected systems. We suggest using more local computing resources, like the fog servers, for these tasks in real-time. Our method carefully balances the real-time workload limits with the higher data transfer delays and extra costs of using cloud services by formulating the difference equation to obtain steady-state solutions for our proposed framework. Our offloading rule checks many system metrics to select the optimal resources. It is focused on the amount of tasks a resource can execute compared with the costs of processing them on fog and cloud servers. We carefully check our method, considering varying task sizes with both fog and cloud servers. Our proposed framework, which includes the waiting time for task queuing based on the M/M/1 approach, offers a dynamic and cost-aware resource selection method to remove data from IoT systems that connect with cloud-fog networks. The proposed approach was observed to manage workloads well and improve processing costs substantially.