Cloud and edge computing enable industrial companies to find solutions for industrial IoT, data analysis and digital business models. Both technologies are growing together more and more and form the new model for industrial IT. But what does this future look like in concrete terms?
Edge computing has become an indispensable part of applications for the Industrial IoT (Internet of Things). Many digital business models require using edge computing to collect and analyze all data in a meaningful way. The cloud alone is not enough because the data is very extensive in many cases. When thousands of sensors are bundled, the bandwidth of the available Internet connection is quickly exhausted. So only aggregated and already partially analyzed data should get into the cloud. The typical hardware for this is microcontrollers or microcomputers like the Raspberry Pi. The costs of the devices are manageable, but their performance is not always sufficient.
Demanding Tasks Require On-Site Computing Capacity
As digitization matures, companies want to perform more demanding tasks with edge computing and the industrial IoT. A typical example is monitoring all types of company premises with cameras and suitable AI applications that recognize dangerous situations, for example. But here, a problem arises: cameras with higher resolution generate enormous data streams – with an 8K camera that can be 100 Mbit per second and camera. Sending such data to the cloud is unrealistic given the limited bandwidth, especially in situations requiring rapid evaluation. Such a situation could, for example, be the automatic alerting of supervisory staff in a train station as soon as a person steps onto the track.
For this, the automated evaluation of the images must take place on-site – and powerful edge servers can make this possible today. With their high computing capacity, they can evaluate video streams in real-time and only send parts to the cloud – for example, as a reference when a dangerous situation has been detected. The AI application can then immediately trigger an alarm based on this.
Edge Computing Allows The Broad Use Of AI
In addition to the limited bandwidth, cloud connections’ high latency (distance-related time delay) also causes difficulties. Numerous industrial applications require short response times. An evaluation of data in the cloud cannot guarantee this. One example of this is the monitoring of machines and systems with machine learning (ML): the system uses historical data to learn the signs of an impending malfunction. As soon as these values are recognized, the system must react and alert the operating personnel.
For such ML models to work correctly, they must be trained with large amounts of data. A dichotomy has been established: cloud computing helps with considerable resources in training the models, while edge computing executes the trained models. The latter is less memory-hungry than the training systems. This enables users to achieve the quick reactions that make many AI applications possible in the first place. In short: Edge computing makes AI easier and more widely applicable. The development in the companies points in this direction. According to recent studies , 73 percent of large and almost 60 percent of smaller companies already use machine learning.
Digital twins are similarly widespread. They map the actual production processes on virtual digital models and feed them with data in real-time. The area of application is monitoring, control and simulation. In this way, a digital twin can easily demonstrate how changes in the processes would affect and enable immediate decisions thanks to the real-time data.
Edge And Cloud Work Together
Many analysts assume that edge computing will develop into an exponentially growing market through such application areas.
The Edge, therefore, has great future potential, and it has been shown that more and more time-critical tasks of the cloud are migrating there. This is how parts of IoT stacks are executed in the Edge. The hyperscale’s and many smaller service providers expand their cloud offerings to include edge services. The customers use so-called “edge appliances”, operated at their location and remotely managed by the service provider.
This collaboration between Edge and the cloud will soon become the norm in industrial IT. But that brings new security requirements for companies because there is no longer a strict separation between IT and operational technology (OT). All edge devices and the machines and systems networked via them are provided with certified device identities and can thus be unequivocally recognized and given access rights to central resources. At the same time, companies are introducing new security concepts, such as the “Zero Trust” paradigm. It dispenses with assumptions such as “Everything is safe in the LAN” or “The danger comes from outside”. Instead, it relies on mutual authentication and verification of the identity and integrity of all devices. Key technologies for IT security such as IAM (Identity & Access Management) or the issuing of certificates to identify the devices are used for this. According to an IDG study , 38 percent of the companies surveyed already use a zero trust model, and 41 percent are currently implementing it.
A Cloud Ecosystem For More Data Sovereignty
From the point of view of many companies, the topic of security also includes the topic of data sovereignty. This is understood to mean the greatest possible control over one’s data, for example, to protect it from access by foreign governments. The relevant example: The state organs of the USA require access to the data in non-US branches during investigations in US companies. Theoretically, this would also be possible in the Edge via the detour of appliances from the hyperscale’s.
For their data sovereignty, companies should rely on a solution for all personal data that meets two criteria: On the one hand, the standardized criteria of data protection – this is confirmed by audits and certifications such as SOC-2 or ISO 27001. At the same time, the solution is also subject to the exclusive regime of European laws and regulations. In numerous cases, companies will, therefore (for the time being), opt for a private cloud, although a regularly checked and certified cloud service is often more secure.
One thing is clear: for data-based business models or data platforms, an alternative to the hyper scalers public clouds is required that is similarly functional and powerful. The EU-wide project Gaia-X seeks to develop such an alternative. The aim is not to create a new hyper-scale. Instead, the project envisages a Europe-wide, networked data infrastructure based on open-source software (OSS). Different central and decentralized infrastructures are networked to form a common system. The goal is a trustworthy digital ecosystem of European and global cloud providers and their offers.
With the increasing merging of edge and cloud computing, OSS stacks, and a “data sovereign” cloud ecosystem, all elements are available for successful, highly scalable digital business models. As a result, companies have the guarantee of being able to operate their industrial IT securely, by data protection law and without vendor lock-in.