A simple explanation of edge computing

A popular explanation of edge computing

As the Internet of Things becomes more and more popular, various concepts and technologies come with the Internet of Things, one of which is edge computing. Of course There is also fog computing. In fact, edge computing and fog computing are similar. Fog computing is just relative to cloud computing. It’s just called edge computing, which is relatively high-end.

Now we will talk about edge computing in a popular way.

Why should I speak in a layman’s terms? I’m afraid that if it’s not in a layman’s terms, you won’t understand. When new things come out, they often require a process of acceptance and understanding. Just like when the Internet first came out, many people didn't know about the Internet, so we had to slowly popularize it and let everyone slowly accept and understand it. Who still explains what the Internet is?

Edge computing has been around for some time, but with the development of the Internet of Things, the concept of edge computing has also become popular. Let’s first look at a non-popular introduction to the concept of edge computing:

Edge computing is a distributed computing architecture. Under this architecture, the operations of applications, data, and services are moved from the network center node to the logical edge nodes of the network for processing.

In other words, edge computing decomposes large services that are originally completely processed by central nodes, cuts them into smaller and more manageable parts, and distributes them to edge nodes for processing.

Edge nodes are closer to user terminal devices, which can speed up data processing and transmission and reduce delays.

The above is an explanation of edge computing that I excerpted from an online article. The entire explanation is basically in professional terms. If you are engaged in industrial control, after reading this paragraph, tell me what edge computing is.

As a programmer involved in the development of edge computing products, I decided to write an article to explain this edge computing in a popular way.

First of all, I want to give an inappropriate example.

For example, there is an APP. When a user uses this APP, it will collect the user’s information, such as the user’s age, gender, mobile phone number, address location, search history, etc., and The main purpose of collecting this information is to better analyze the user's behavior and interests, such as cars, houses, books, food, etc. Then deliver content and ads to them more accurately.

This is a very common function, but for such a function, how can it be linked to edge computing.

Before edge computing, there was cloud computing.

If cloud computing is used, the behavior of this APP is as follows:

After the APP collects the information, it uploads all the basic information to the server, and then the server Execute algorithms to calculate and identify the user's interests and hobbies, and may even calculate the user's spending power. The server can then deliver content and advertisements that the user is interested in based on this calculated result.

If edge computing is used, the behavior of this APP is like this:

After the APP collects the information, it does not upload it to the server. Then the APP itself calculates and identifies the user's interests and hobbies, and can also calculate the user's spending power, which is the computing function of the server, which is directly completed by the APP. Then the server only needs to ask the APP which user is likely to have an annual salary of one million and which user is single. The APP only needs to tell the server that the user All the Way East is handsome, single, likes traveling and writing poetry, and can provide dating content for beautiful girls.

In this way, the server does not participate in the calculation during the entire process, and the server does not participate in collecting information. Because this information is collected and calculated in the APP itself and is not uploaded, there is no information collection involved.

And, this is edge computing.

That is to say, the part that was previously calculated by the server is now directly calculated by the information collection device, and then the calculation results are directly output to the server. The server only needs results and does not need process data.

Let’s talk about this edge computing in a popular way in the form of answering questions.

So, what is edge computing.

To put it bluntly, edge computing means that (server) cloud computing is too lazy to calculate. For this little data, you can do it by yourself when collecting data. It is very tiring to throw everything to the server. of. So, edge computing came here.

So, where are edge computing used in the industrial control field?

There are too many. As many PLCs, controllers and touch screens are beginning to be connected to the Internet of Things, each device needs to collect different information, including temperature, humidity, output, production data, operating status, etc. The parameter indicators and performance data of different industries are different, which makes it difficult to form a set of standards on servers through cloud computing. This makes PLCs, controllers, etc. all use edge computing.

Why were edge computing not popular with DTU or IoT modules in the past, but now they are becoming popular?

Because the processing capabilities of the modules or chips used in the current IoT are getting higher and higher, and the resources are relatively abundant. With the decline in the cost of some chips and the simplification of the development model, some chips or modules are After dealing with the basic data collection functions, there are still excess resources and low function utilization. That is, if you only use 10% of a 100% chip or module to collect data, you still have 90% that you can use. Computing

So, what are the advantages of using edge computing.

1 can increase the number of supported devices by several orders of magnitude.

For example, a server has 10,000 health points. To connect a device, 1 blood point will be consumed. If the device is analyzed for data, 9 blood points will be consumed. In other words, it takes 10 blood points to connect and calculate a device. Then this server can only connect to a maximum of 1,000 devices before it hangs up.

If the server is only responsible for accessing the device and does not perform calculations and analysis, then connecting to a device consumes 1 blood point, and the device itself performs data calculation and analysis, and then outputs the results. At this time, the server can access 10,000 devices.

? Without edge computing, the server can connect to 1,000 devices.

? If edge computing is used, the server can connect to 10,000 devices. Improved by an order of magnitude. For some complex equipment, especially some factories, on-site operations, etc. that require a large amount of data, this advantage can be even more reflected if edge computing is used to save space and resources for the server.

2 Make computing more flexible and controllable

As mentioned earlier, it is difficult for servers connected to devices to achieve unified computing and analysis standards, because the Internet of Things is a universe of all things. The data collected by each device in the connected network is different. If edge computing is used, corresponding calculations and analyzes can be performed individually for each device. Of course, if the same equipment or parameters are used, they can be copied using the same set of calculation standards or algorithms. If the calculation script is opened to users, users can customize and add their own calculation formulas and behaviors.

What are the modes and topologies of edge computing?

For example, in a data collection system, with a cloud server as the center, mobile clients, PC clients or third-party interfaces are connected to the cloud server to obtain data, and the data collector, The data collection module is used to connect to the cloud service.

The data acquisition module can collect PLC, frequency converter, smart instrument, etc., upload the data to the cloud server, and the server will perform data analysis and calculation, and then PC or mobile client, third-party interface can obtain it Results of data analysis. However, in this case, as more and more devices are connected, the burden on the cloud server will become heavier and heavier, and more and more types of PLCs, controllers, etc. are connected. The original cloud service data The computing model is difficult to meet increasingly complex applications. At this time edge computing comes into being.

Edge computing can be seamlessly introduced while the original topology remains unchanged. Open the edge computing function on the data collection module side, and perform complex calculations, policies, rules, etc. by the data collection module. After obtaining the output results, you only need to upload the results to the cloud service. Then the PC client, mobile client and third-party interface obtain it from the cloud service.

For example, the data acquisition module needs to collect an electric meter. The data that the electric meter can collect include current and voltage, but not power. Of course, current electricity meters cannot collect very little power. This is just an example.

So what should we do? Customers really want to see the power. When there is no edge computing, in order to see the power, we have to add certain calculation rules in the cloud service and calculate the collected current and voltage to obtain the power. If there are 1,000 electricity meters, the cloud server will calculate these 1,000 electricity meters. This increases the workload and burden of the cloud server.

If there is edge computing, then the calculation function can be added to the data acquisition module, and the power can be obtained by directly calculating the collected current and voltage, and only need to upload the power to the server. In this way, even if there are 50,000 electricity meters, the cloud service will have no computing pressure because it does not require calculations.

This is a popular talk about edge computing.