New decentralization paradigm

In edge-to-cloud architectures, the edge and cloud have different roles: The edge is used for real-time analysis and action, while the main role of the cloud is correlation and coordination. Correlation of relevant data for the purpose of deep learning and coordination of company-wide processes. © HPE

This points to a new IT paradigm. After the last wave of centralization, which focused on the cloud, the trend is now moving back towards decentralization: the importance of the periphery, understood as everything that can be equipped with a data-generating sensor, is increasing. Market research company Gartner predicts that in six years' time, three quarters of company data will be generated and processed outside all data centers, at the edge. Today, according to Gartner, this figure is ten percent. Until now, the prevailing view has been that data generated in the Internet of Things (IoT) is processed in the cloud. However, transporting the sensor data there takes far too long and is also too insecure. Delays at the edge can literally have fatal consequences if machines do not switch off or traffic lights do not change. In addition, bandwidth is often scarce or only available in phases at the edge of the network, which can literally be anywhere. And: it is unnecessarily expensive to send masses of mostly redundant data through the network.

The HPE Edgeline OT Link platform includes a no-code workflow generator and application catalog that enables customers to orchestrate components, data and applications through a graphical user interface. The platform integrates an ecosystem of cloud and data center applications - from AWS, Google, Microsoft, SAP and GE, among others - to bring insights from the edge to the entire enterprise and supply chain. © HPE

A new approach is therefore needed for edge systems to fulfill their role as a bridgehead for OT-IT integration. A model for this can be found in our pockets: the smartphone with its apps.

Everything from a single source

Older people still remember the days when you had a phone for making calls, a Walkman for listening to music, a camera for taking photos and a PC for surfing the internet. Steve Jobs had three ideas that radically changed this situation. Firstly, he physically integrated the phone, music player, camera and Internet into one device. Secondly, he created an ecosystem of applications that constantly created new ways to creatively utilize these physical components. Thirdly, he made the application so simple that really any fool would be able to use it. This model can be transferred almost one-to-one to the edge systems that form the core of cyber-physical systems. Private life becomes easier when you only have to operate one device rather than five different ones. OT-IT integration also simplifies processes in industrial environments.

In automotive manufacturing, for example, different test and measurement systems are often used for production-synchronous testing and engineering. One car manufacturer decided to replace this heterogeneous environment with edge IT systems with an integrated PXI interface module. With a single device class, he was able to process large volumes of vehicle data in the desired level of detail in real time and set up applications such as hardware-in-the-loop simulation, rapid control prototyping, bus monitoring and automation control. This standardization enabled the car manufacturer to achieve such great efficiency gains that it was able to increase the number of vehicles produced per day.

Despite all standardization efforts, industrial environments are still characterized by great heterogeneity. Convergent edge systems must therefore offer a wide range of options for controlling OT systems as standard - such as digital I/O, CAN bus, Modbus or Profinet - either via a multi-protocol standard such as PXI or via dedicated adapters. Programming interfaces and field programmable gate arrays (FPGA) also allow the customized creation of OT adapters, including the implementation of programmable logic controllers on the Edge IT system.

Application development without coding

As with smartphones, the true value of convergent edge systems only becomes apparent with the applications that link the various OT and IT components into a useful process. For example, video quality control in production: a video camera records the products rolling past on a conveyor belt; a machine learning program uses the video recordings to detect product defects; a detected product defect triggers an impulse in a programmable logic controller to push the defective product off the conveyor belt.

In many cases today, this kind of organized interaction between OT and IT is set up and programmed manually - which involves a great deal of effort and is prone to security vulnerabilities. One solution to this problem is so-called no-code workflow generators. The data sources, adapters, drivers, middleware, applications and PLCs can be combined into a workflow by linking corresponding graphical symbols with the mouse. The IT applications with which the OT data is analyzed and processed - such as PTC ThingWorx or SparkCognition - are selected from an application catalog and, packaged in a container, transferred from the cloud to the edge system.

Foolproof system management

Once all these requirements have been met, the problem of system management at the edge still needs to be solved. Administrators are not dealing with hundreds or thousands of systems in the air-conditioned and access-protected rooms of their local data center - but with hundreds of thousands of systems in all kinds of remote and inhospitable locations. Setting up and maintaining the systems manually is therefore out of the question. Instead, what is needed is a system management system that - similar to a smartphone - performs most tasks autonomously. This includes a wealth of tasks such as commissioning the edge hardware, loading and installing firmware, operating system and applications, keeping them up to date, system monitoring and troubleshooting.

Other edge-specific conditions that the system management must take into account include a fragile network connection - typical in mobile deployment scenarios or at remote locations. This means, for example, that interrupted update processes are resumed at the point where they were interrupted - and that the edge system actively reports when it is back online after a network interruption. In addition, the system management must offer edge-specific security functions because there is no access control at the edge as there is in the data center.

Factory revolution

The invention of the iPhone triggered a revolution that has changed the way we live and do business. The same must happen with edge IT systems, which, according to Gartner, will bear the brunt of the digital universe in the future. Only then can the rosy promises of efficiency, growth and revenue of the Industrial Internet of Things be realized. Physical OT-IT convergence, no-code application development and foolproof system management at the edge are the keys to this. The convergence of OT and IT will shift the coordinate systems of these industries.

OT and IT manufacturers are conquering foreign territories and forging new partnerships to create new services, products and business areas - at the same time, they have to defend their traditional territories against attacks. Vendor lock-in in the OT sector today is reminiscent of the situation on the mainframe market 30 years ago due to the many proprietary technologies. The world of IT has made the arduous journey from proprietary to open systems. The world of OT has yet to do so.

Peter Widmer, Global Category Manager Converged Servers, Edge & IoT Systems at Hewlett Packard Enterprise