Recently, the term “fog computing” has come to be used in conjunction with cloud computing and edge computing. “Fog computing” is a relatively new term proposed by Cisco Systems and is often used in the same context as edge computing. So why did we need to create a new word? Here, we will explain the meaning of “fog”, the difference between fog computing and edge computing, and the “distributed processing” common to fog computing and edge computing. We will also discuss the role of each server in a typical cloud fog/edge physical device three-tier model.
What is “fog”?
What does fog mean? Fog is fog in English. The word “cloud” in cloud computing, which has become a common term today, means “cloud”. In other words, it means that it is closer to the ground than the cloud. For example, if a physical device such as a sensor is “ground”, it is a general term for systems that exist closer to the cloud than clouds and that hold between the cloud and the physical device. In addition, the word “fog” may have the meaning of dispersing clouds to make them look like fog.
A commonly used and similar term for a system that runs between the cloud and physical devices is “edge computing.” In fact, edge computing and fog computing are not used in a very strict way, and it can be said that they are used with almost the same meaning at present. There are many examples of writing “edge/fog computing” together.
The difference between “fog” and “edge”
So why did you dare to create a new word? With this in mind, the difference between edge computing and fog computing becomes apparent.
Edge computing is a technology that reduces the load on the server by processing data in a location close to the user and not focusing on the cloud. So again, it does the work between the cloud and the physical device, except that edge computing, is closer to the physical device, while fog computing is a bit closer to the cloud. Fog computing aims to optimize resources while applying cloud-side technology to edge servers, and performing distributed processing in near real-time time. So to speak, the goal is to get closer to cloud computing while preserving the capabilities and benefits of edge computing.
Conversely, by moving the functions of the cloud server to the edge server, it can be said that the processing on the cloud server is distributed. As mentioned earlier, you can think of the term fog computing as coined by comparing the decentralization of cloud servers to the dispersal of clouds into the fog.
The need for distributed processing
What is common to edge computing and fog computing is “distributed processing”. Cloud computing, on the other hand, is “centralized processing” in principle. It can be said that this “distributed processing” is the reason why edge computing and fog computing are in the limelight. So why is distributed processing needed?
One of the backgrounds is the limitation of cloud computing. The disadvantages of cloud computing, or centralized processing, are that the amount of data sent to the network is large, which slows down the speed, and the data is concentrated on the cloud server, leaving security concerns. Therefore, it becomes necessary to set up an intermediate server between the physical device and the cloud server, and perform distributed processing here. Depending on whether this intermediate server is closer to the physical device (edge server) or compatible with the cloud (called fog server for convenience), depends on whether it is called edge computing or fog computing.
In the above example, we considered a three-layer model of a cloud server, intermediate server, and physical device, but in reality, there is no cloud server, and it is composed of two layers of multiple servers and physical devices. There is also a system. In such a system, the processing performed by the cloud server in the three-tier model is processed by the server in front of the cloud. This group of servers also includes edge servers, but it is the closest to the system that fog computing is aiming for in that it has both cloud server and edge server functions.
An example of the role of each server in a three-tier model
Finally, let’s review the role of each server in the three-tier model. As mentioned above, it may be a two-tier model, but here we will consider a three-tier model of a general cloud server, intermediate server, and physical device.
First, the role of the edge server is to collect data and respond to emergencies. Emergency response is, for example, a temporary emergency stop of a robot when it is likely to hit a person. Edge servers are basically faster in terms of being closer to where the data is generated, so edge computing is probably the most suitable for urgent processing. Therefore, artificial intelligence (AI) used in applications such as predictive maintenance will basically be placed on the edge server because of the demand for speed.
The role of the fog server is to store and analyze data and send the results to the cloud. In some cases, the fog server may collect data directly or act as a cloud server on your behalf. In the artificial intelligence example given above, it could be placed on a fog server for applications where response speed is not required.
The cloud server is responsible for managing the data from the edge server and fog server, displaying it to the person in charge, and sending the analysis results to a higher-level system (for example, a business management system). Also, in edge computing, cloud servers will often store and analyze data.
Things to watch out for when deploying edge/fog computing
This time, I took a brief look at the basics of fog computing. The construction of edge computing and fog computing will need to be done based on the strategy of the production system as a whole, while carefully considering the situation in the field. In addition, it cannot be said that this is generally good in terms of which process is processed by which server and a sense of balance are required. Edge computing and fog computing can be said to be fielded in the future, so when asking an outside specialist, I would like to pay attention to these points and proceed with the construction of the optimum system.
