"Edge computing" can be translated as the mathematical processing of data at the network edge. This means that business processing is done close to its data sources. For example, it allows Industrial PCs to filter incoming data via various data acquisition tools before transferring this information to the cloud.

In 2021, technological advances enable faster data transfers, meanwhile, the growing number of connected objects drastically increases the volume of data to be shared. It is estimated that these data volumes would be too large to transfer in raw form and wait for business processing in the cloud. This is where Edge Computing comes into play and allows for more and more "real-time" processing locally (autonomous cars, connected watches with immediate diagnostics, or production lines with artificial intelligence…).

Today, we process the information collected through data acquisition devices in datacenters or other Cloud infrastructures, it is the classic method. This means that for a factory that collects data on machine temperatures, reports and interpretations of the heat curves of the devices are made on remote servers. The data must therefore transit through the network to be transformed and then used (we find the famous ETL, Extract  -> Transform -> Load). Edge Computing is an architecture that brings more lightness to this process since instead of sending the data to a datacenter, it is a computer or server closer to the source that performs these calculations. 

To illustrate this point, Apple users are also users of Edge Computing, perhaps without knowing it. The facial recognition of smartphones collects the data and then processes it using a machine learning algorithm and ultimately provides a decision directly to the user (unlocking or not unlocking the phone) all done locally within the iPhone.

Ultimately, the principle is not new, the evolution lies in the ability of end devices to be both collectors and processors.

There is significant use of Edge Computing in the context of the Internet of Things (IoT), and this is also true for its application in industry (IIoT). Connected devices collect a large volume of data, so it is preferable to filter them before transferring to the Cloud or a data center. 

• One of the primary advantages of this technology is a certain reduction in bandwidth between end devices and storage servers.  
• In real-time use cases, Edge Computing makes perfect sense as the observed latency is significantly reduced for data processing. Overall, we reduce the data in transit over the network and improve the performance of the endpoints.
• Additionally, sensitive data can remain stored locally, and in the event of a server compromise, the amount of data at risk is reduced. Nevertheless, while it is generally accepted that sending less data over the network is more secure, nodes at the network's edge are potentially more vulnerable than large data centers. Therefore, it is essential to secure exchanges as much as possible: data encryption, use of VPNs, access control…

In our modern context, the evolution of data transport means, particularly the arrival of 5G, marks a new era in the collection and processing of information on the internet. The use of Edge Computing can be very relevant in certain real-time contexts; however, one must remain cautious and consider the limits of such an architecture. First, the more powerful the terminals are and the more capable they are of performing calculations, the higher their costs rise; the budget is a very important factor for any project. Additionally, data security is becoming increasingly concerning these days, so be careful not to overlook this point in this type of architecture. Finally, implementing a network architecture like this is complex and requires confirmed technical skills.