A Dynamic Algorithm for Fog Computing Data Processing Decision Optimization

Continuing Cloud service challenges like vendor lock-in and security in addition the Cloud clients' increased reliance on the service are motivating large Cloud tenants to build their own Clouds. Cloud providers still have the advantage (cost, availability and expertise) and therefore, a hybrid approach is still the first choice for Cloud clients aspiring to achieve partial independence. This promotes utilizing Fog computing architectures where resources rented from Cloud providers are supplemented by nodes on the edge of network that have limited computational capacity but are closer to request source. Fog nodes offer an opportunity to filter and process requests on the edge of the network in order to distribute the load and minimize the network congestion caused by low-value data. A question is, in turn, posed to Cloud architects on how to optimize the decision as to where to perform each step of the request service. The aim is to mitigate both the risk of pushing high loads to the Cloud servers which increases the cost; and the risk of localizing the whole process and losing the benefits from Cloud services. We build upon our previous work to solve this problem by proposing a novel dynamic programming algorithm to optimize the data processing decision for requests in a Fog environment. The dynamic programming algorithm achieves closer results to the optimal solution by reaching optimal increments of solutions from previous steps in feasible time. Initial experimental results comparing 5 heuristic algorithms are presented with the purpose of offering insight into contradicting factors impacting the problem (cost, network capacity, node and Cloud capacity).