Читать книгу Fog Computing - Группа авторов - Страница 47

In general, fog servers have limited resources in which they can serve a fairly limited number of tenant-side applications at each time slot. Therefore, fog servers require dynamic and optimal mechanisms to support their serviceability. Here, we list the basic elements involved in the operation management of fog servers.

 Load balancing of request and traffic. Commonly, fog servers are connecting with one another vertically or horizontally. Therefore, it is possible to establish a cluster computing group among the fog nodes connected in 0-hop range toward enhancing the overall computational capability. Besides the computation-related loads, since fog nodes are fundamentally Internet gateway devices, the heavy network traffic can always affect their serviceability. In order to overcome the traffic-related issues of fog servers, the provider may configure multilayered caching mechanism that utilizes the fog nodes in the hierarchy to reduce the burden [66].

 Server allocation, server scheduling, and server migration. Three corelated elements, especially for the resource constraint mobile fog node (constraint mFog), such as UE-fog and UAV-fog nodes. To explain, a provider may deploy a specific type of fog server on the constraint mFog device for a domain-specific application in a specific period of time. Whereas, the rest of the time, the device does not operate the fog server at all. For example, in an indie fog environment [17], the owner of a smartphone (UE)-fog may configure the device to serve the context reasoning–based fog server [36] only when the owner is carrying the device in outdoor areas and the battery level of the device is over 50%. Further, the owner can also configure that, when the battery level of the device is between 51 and 70%, it will redirect/migrate the request to another authorized fog node. Similarly, the notion described here is applicable to other MFC domains, such as LV-fog [24] and UAV-fog [63].