Query Latency Optimization by Resource-Aware Task Placement in Fog

Abstract

The advancement of IoT (Internet of Things) technology has led to the proliferation of IoT-enabled applications. These IoT applications demand low query latency for fast data analytics. Fog computing has aided in reducing the query response time, but challenges still exist regarding query latency reduction in network-compute heterogeneous fog environment. In this paper, we propose a query latency reduction approach that formulates the query execution plan in a network-compute aware manner by considering the resource capacity of fog nodes and current network conditions. We introduce a query task placement algorithm that performs task placement by jointly considering both compute and network resources. The proposed algorithm selects set of nodes for query task placement based on minimum-latency criteria. Moreover, the proposed algorithm mitigates the computational bottleneck by offloading the tasks of computationally overloaded nodes. The proposed approach reduces latency by 71% and 24%, and decreases network usage by 52% and 35% compared to other approaches.