Technologists and analysts agree that Edge Computing will be increasingly important for manufacturing companies that want to make decisions based on data generated in production.
Decentralize computing power, placing it closer to where data is generated. It is one of the technological trends on which experts and analysts are now in agreement. And even companies that have embarked on a digitization journey are realizing that, with the exponential increase in the volume and speed with which data is generated, transmitting all this information in a cloud or a central data center is neither effective nor efficient.
“Currently, about 10 percent of enterprise-generated data is created and processed outside a traditional centralized data center or the cloud,” said Santhosh Rao, senior director of research at Gartner. “By 2025, this figure will reach 75%.”
Edge vs Cloud?
After a few years in which there was only talk of the cloud, a new technological trend is therefore on the way. However, let's face it right away, it is not an alternative to the cloud, but a complement. Let's see why.
First: even with the arrival of 5G and its characteristics so close to the needs of the industrial world, connectivity still represents a bottleneck and a critical point ("single point of failure", as the Americans like to say) for the industry: supervision, whether it is a machine, a line or a plant, for some processes is too important to risk losing it and the only solution for now is still to have it reside in situ, on the plant.
Second. The amount of data that is generated by a production process can hardly be historicized and managed completely in the cloud. And if it is no longer so much a performance problem, there is still the cost aspect: both the storage and the subsequent analysis of industrial big data would require space and computing power disproportionate to the economic resources that could be put in place small and medium-sized enterprises (but also a large part of large multinationals).
In short, thinking of using the cloud to perform functions that are not within its capabilities would be a mistake. The cloud, on the other hand, remains a more than valid solution for second-level analyzes on data that has already been processed, not raw (the raw data of which we know that industrial plants are large producers), very useful for comparing series of data from different locations or departments to draw potentially interesting correlations.
Having clarified that the collection and initial processing of data must take place in a place that is close to the data sources, the point is to identify where. An intermediate step is necessary between the production plants and the enterprise network. A level that is, in fact, at the "periphery" of the system (the edge), where not simple "little boxes" that simply act as a bridge can be put to work, but high-level devices that can give solid certainties.
What features should an industrial edge computer have?
Having clarified that edge computers perform a function of crucial importance in the creation of value in manufacturing companies, let's now see what characteristics a solution that is technically adequate for the purpose must have.
First of all they are needed high performance as regards computing and storage capacity: these are devices on which we must be able to "host" both the supervision system and the process database. If we succeed, then we will be able to send data already "processed" and useful for advanced analysis to the cloud.
Second, you need a solution that provides at least high availability, if not fault tolerance. We are talking, that is, of a level of redundancy that guarantees a uptime (service time) greater than 99,99%, thus keeping supervision available (and its visualization via client) practically always.
As a corollary of this feature, edge computers are required to be easily replaceable (and obviously have a systematic backup of the applications used).
Third, you need to enable a secure connection to the cloud: the devices chosen to govern the factory or system data must let only authorized data and commands go out (and enter). Therefore they must include an active filter, both incoming and outgoing, useful for protecting both plants and cloud architectures.
Fourth, edge computers must represent where fame occurs convergence across the IT domain (Information Technology) and OT (Operational Technology), i.e. between the world of information technology and that of production. A meeting point that must be reliable and easily manageable: IT, in fact, is already familiar with the Edge concept, but often lacks knowledge regarding the processes and devices that are connected downstream (PLC, SCADA and control systems industry in general).
By choosing the right solution for industrial edge computing - where "right" means a solution created and designed for this purpose - manufacturing companies and utilities will be able to achieve the goal that underpins every choice of technological advancement: Making meaningful information available to management on the basis of which to make business decisions.