Joint Task Segmentation and Server Selection for Real-time Stream Data Analytics in Edge-cloud Environments

To reduce the task completion latency and wide area network traffic for real-time stream data analytics applications, edge computing, which pushes computing resources from the remote cloud to the network edge, has drawn both researchers’ and engineers’ attention. Since edge servers are typically resource-limited while the tasks are typically computing-consuming, a practical deployment method is to split the task into two sequential parts and to execute them on the edge and the cloud separately. Combing it with the servers’ heterogeneity in terms of computation, memory, etc., we are faced with the joint task segmentation and server selection problem. To this end, this paper formulates it with the goal of minimizing the long-term energy cost while meeting the real-time data processing constraint. To solve the problem, we design an online randomized iterative algorithm, which is based on the Lyapunov optimization framework and the Markov approximation method. Finally, trace-driven simulations are conducted to evaluate the performance of the proposed algorithm and the result shows its superiority over the heuristic alternatives.