4 Architecting IoT Cloud
239
Despite the above definition from Cisco, many companies feel that fog and
edge computing are basically the same because both are focused on utilizing local
network computing capabilities to complete computing tasks normally completed
by the Cloud. In this context, several companies suggest that the biggest difference
between fog and edge computing is where the data processing occurs. Edge
computing generally takes place on the actual endpoint IoT devices (IoT things)
or on a gateway device that is physically close to sensors/actuators. However, fog
computing pushes edge computing to data processing centers connected to the LAN
or to actual LAN hardware that is more physically distant from actuators or sensors.
Key features of fog and edge computing include [34, 35]:
• Heterogeneity – There is a wide heterogeneity of IoT edge devices (e.g., vibration
sensor, temperature sensor), there is a wide heterogeneity in communication
technologies and protocols (e.g., OPC-UA, Modbus, CAN bus, BACnet, MQTT,
REST, SICK SOPAS), and there is a wide heterogeneity in network technologies
(e.g., Wi-Fi, LTE, 3G/4G, Bluetooth) as well. Fog acts as a multi-protocol building block that can be utilized in diverse environments for protocol translation,
flexible integration, data delivery, and device management.
• Interoperability – Fog should be integrated into many solutions in order to
support a broad range of different services such as data streaming.
• Geographical Distribution – Fog computing is deployed in a distributed manner
to provide top-quality services for stationary and mobile end devices.
• Edge Processing/Storage – A wide range of applications can be executed on the
edge node close to the data source in order to reduce response time and save the
bandwidth between edge and Cloud. Edge can also be utilized as a short-time
historical storage.
• Quality of Service – Fog computing emerged partly to address quality-of-service
constraints of IoT endpoints. To name a few, real-time video streaming, gaming,
and CCTV monitoring are among those applications that demand low-latency
services.
• Real-Time Interaction – Fog can be used in real-time applications, including
real-time traffic monitoring, require real-time processing speed, and capability
as opposed to batch processing.
• Large-scale Sensor Networks – Fog computing is very useful to be utilized in
large-scale sensor networks (e.g., in smart grid or for environmental monitoring
applications) in which utilizing systems with distributed storage and computing
resources are required.
As noted, Cloud computing in IoT networks offers several benefits including
exceptional computing efficiency, enormous storage capability, and wide-area
coverage. On the other hand, edge computing offers a device-centered process,
increased mobility, high QoS, resource pooling at the edge, and the ability to manage
data in real time. Table 4.8 presents the main differences between Cloud and edge
computing.
239
Despite the above definition from Cisco, many companies feel that fog and
edge computing are basically the same because both are focused on utilizing local
network computing capabilities to complete computing tasks normally completed
by the Cloud. In this context, several companies suggest that the biggest difference
between fog and edge computing is where the data processing occurs. Edge
computing generally takes place on the actual endpoint IoT devices (IoT things)
or on a gateway device that is physically close to sensors/actuators. However, fog
computing pushes edge computing to data processing centers connected to the LAN
or to actual LAN hardware that is more physically distant from actuators or sensors.
Key features of fog and edge computing include [34, 35]:
• Heterogeneity – There is a wide heterogeneity of IoT edge devices (e.g., vibration
sensor, temperature sensor), there is a wide heterogeneity in communication
technologies and protocols (e.g., OPC-UA, Modbus, CAN bus, BACnet, MQTT,
REST, SICK SOPAS), and there is a wide heterogeneity in network technologies
(e.g., Wi-Fi, LTE, 3G/4G, Bluetooth) as well. Fog acts as a multi-protocol building block that can be utilized in diverse environments for protocol translation,
flexible integration, data delivery, and device management.
• Interoperability – Fog should be integrated into many solutions in order to
support a broad range of different services such as data streaming.
• Geographical Distribution – Fog computing is deployed in a distributed manner
to provide top-quality services for stationary and mobile end devices.
• Edge Processing/Storage – A wide range of applications can be executed on the
edge node close to the data source in order to reduce response time and save the
bandwidth between edge and Cloud. Edge can also be utilized as a short-time
historical storage.
• Quality of Service – Fog computing emerged partly to address quality-of-service
constraints of IoT endpoints. To name a few, real-time video streaming, gaming,
and CCTV monitoring are among those applications that demand low-latency
services.
• Real-Time Interaction – Fog can be used in real-time applications, including
real-time traffic monitoring, require real-time processing speed, and capability
as opposed to batch processing.
• Large-scale Sensor Networks – Fog computing is very useful to be utilized in
large-scale sensor networks (e.g., in smart grid or for environmental monitoring
applications) in which utilizing systems with distributed storage and computing
resources are required.
As noted, Cloud computing in IoT networks offers several benefits including
exceptional computing efficiency, enormous storage capability, and wide-area
coverage. On the other hand, edge computing offers a device-centered process,
increased mobility, high QoS, resource pooling at the edge, and the ability to manage
data in real time. Table 4.8 presents the main differences between Cloud and edge
computing.