Delay-aware resource allocation for data analysis in cloud-edge system

There is a strong need for data analysis in information systems to support various services. Traditional cloud data centers provide powerful ability to conduct data analysis jobs. However, the data transmission consumes a large amount of time and leads to a long service delay. The QoS (Quality of Service) caused by long service delay is unacceptable for real-time services or applications. The collaboration with edge computing is an opportunity for service delay reduction. In this paper, we investigate the task placement problem for reducing service delay in cloud-edge system. We use the W-DAG (Weighted Directed Acyclic Graph) to model the data-intensive service or business logic. We analyze the data and resource requirements for the tasks, which constitute the integrated service, and make resource allocation between cloud data center and edge nodes. Then, we propose the task placement algorithm to achieve shorter service delay. The core idea is to make a tradeoff between data transmission time and data analysis time. The simulation results show that our algorithm has significant performance improvement on service delay reduction.