Workload Allocation Between Edge and Cloud: Delay Minimization, Pricing, and Free Service

Besides adding powerful cloud computers, it is now possible to improve the quality of digital services by incorporating edge computing, an emerging computing paradigm. This prompts organizations to consider an integrated model that augments existing cloud computing systems with edge computing. However, this model requires an appropriate workload allocation between edge and cloud, which makes it challenging to make associated capacity and strategy decisions. To address this issue, we consider a scenario in which a vendor operates an integrated edge-cloud service system. Using game-theoretic analysis, we study the three-way interplay among the capacity of edge computing, the workload allocation between edge and cloud, and the strategy for pricing and service offering. We show that minimizing the expected delay experienced by users is often optimal or near-optimal under full market coverage, regardless of revenue models. Based on this finding, there exists a progression of workload allocation: the workload is processed by cloud computing only, an edge-cloud combination, and edge computing only as the edge computing capacity expands. We also explain how the cost structure of computing options can shape this progression. Other than workload allocation, we discover that the impact of usage on profit is non-monotone and subject to critical thresholds. We also observe that vendors benefit from deploying edge capacity close to customers in a large market. Building on these results, we delineate the impact of adding an ad-supported free service and demonstrate its attractiveness. Our findings unveil how edge computing could shape computing services and may guide vendors' pricing and capacity decisions in various business and technical environments.