Low latency, high connectivity, and efficient bandwidth utilization: Edge computing is ideally suited for IoT applications that require real-time data to be processed as quickly as possible. It offers the best basis for relieving central IT infrastructures of the rapidly increasing volume of data in the future.
Global data traffic is exploding: As the world becomes smarter and more connected, experts expect exponential growth in global data volumes in the coming years. An estimate by the market research company IDC suggests that the amount of data generated could rise to 175 zettabytes by 2025. The Internet of Things (IoT) is becoming a driving factor behind this development: The boom in smart homes, connected production and logistics processes, and IoT projects in public spaces will generate more and more data in future. While the coronavirus crisis has temporarily slowed the growth of the IoT market, analysts expect the sector to recover quickly. In 2019, the global number of connected IoT devices stood at 7.74 billion, according to Transforma Insights, but by 2025 the research firm expects this number to reach 16.44 billion devices.. Some analysts are even predicting double that amount in the same period.
If central cloud infrastructures are left to process the real-time data that will be generated in the future on their own, they will reach their limits sooner or later – the growth involved is simply too rapid for them to keep up. Much of this data, however, would not actually need to be stored in the large data centers, but would be much better off at its actual place of origin – the edge. Decentralized edge computing infrastructures at these locations could analyze and store the data in real time, bypassing any physical detours. After all, the cloud often only needs a fraction of the information collected, especially when it comes to IoT. Consider the example of smart camera systems for traffic analysis: Although the livestream of an intersection generates large amounts of real-time data within a very short time, only specific events are actually of interest to the machine learning algorithm in the cloud. But if the camera system and the cloud then nevertheless exchange countless hours of video material in which no cars even pass the intersection, this places an unnecessary burden on the company's IT resources.