Unloading and consolidation of computing resources in the environment of fog and boundary computing

Abstract

The article analyzes the management tasks in the fog and edge computing when requests of Internet of Things (IoT) devices are processing. The features of the managed objects are determined: the hardware heterogeneity, the mobility of resources, short response time, resource constraints, energy saving. Based on the review of publications over the past five years, the most urgent management problems in the fog and edge networks have been identified. These tasks include the routing of requests, resources offloading, the control of energy savings. The classification of the reasons for the transfer of tasks from one device to another is given. Examples of researches related to the problem of selecting a resource for service requests are given. Traditional models of resources and algorithms description are supplemented with new characteristics relevant to these networks. The example is an indicator of the instability of the presence of a resource in a fog or edge network. One more example is requests separating into two or more inputs with different characteristics, which is typical for many IoT applications. Power saving is the subject of many studies. Challenges are often a reflection of new realities. For example, the aim can be to distribute requests between the data centers in order to minimize the carbon footprint and to meet the requirements for response time. A specific problem is to attract resources to the common pool. Various algorithms are proposed based on game theory. The future priority tasks are listed in conclusion.