A Software-Defined Approach Sparks Digital Transformation of Industrial Automation
Can something be done to avoid the frustration of upgrades that arrive in the vendors’ own sweet time and to overcome other obstacles?
It’s a pivotal time in the industry as many industrial companies are working towards/undergoing a digital transformation. Industrial companies and manufacturers have historically paid steep prices for automation systems purpose-built to perform a single task and lacking the flexibility to adapt to changing market environments. These proprietary solutions are not designed for interoperability with other products, which locks the buyer into the vendor and restricts component choice.
Working with a single industrial automation supplier may sometimes have benefits, but as technology advances and the marketplace demands and expects greater agility, the drawbacks become readily apparent. Proprietary systems are expensive to purchase (high CapEx) and maintain (high OpEx). Because these systems are developed in low volumes and built with highly specialized components, vendors lack the economies of scale inherent in commercial off-the-shelf (COTS) solutions.
Despite the Open Platform Communications (OPC) standard instituted in the 1990s, which enabled communications among proprietary systems, interoperability remains an issue. And as systems become increasingly interconnected, device and data security become paramount concerns. Security cannot be an afterthought. It needs to be designed in from the ground up, yet automation solution vendors often lack the experience to implement a layered security infrastructure leveraging multiple technologies.
A Cue from Telecom
The big issue, though, is that adding features and upgrading systems is costly and difficult, and usually takes place at the vendor’s pace, constraining the operator from taking advantage of the latest technological advancements and innovations (Figure 1).
Industrial automation developers can take a cue from the experience of the telecom sector. At one time, telecommunications service providers also faced a predominance of proprietary equipment, which was a drag on the industry’s growth. Over a dozen of the world’s largest providers got together to lead the transition to interoperable solutions based on industry-standard servers—an approach called network function virtualization (NFV). After a few short years, telecom equipment vendors were able to offer software-based network functions running on COTS servers, making possible large economies of scale, wider vendor choice, and interoperability—all of which has benefited not only the service providers, but also the end users.
Now, a comparable digital transformation is underway in industrial automation, sparked by software-defined infrastructure and enabled by the IIoT. The premise of software-defined infrastructure is that most operations and control functions in an automation system can be consolidated onto standard, high-volume COTS servers capable of satisfying the real-time performance requirements of industrial environments. This creates an efficient, flexible and light-footprint alternative to proprietary industrial solutions. Software-defined infrastructure utilizes open standards and open platforms, extending them to meet industrial requirements, thereby reducing OpEx and CapEx and reaping the benefits of the IT cloud.
A software-defined infrastructure approach allows users, software vendors, and systems integrators to more easily develop interoperable components than proprietary solutions allow. Since software and server hardware are decoupled, software can be easily migrated and reused. Moreover, because the primary hardware platform is a server, it takes less effort to secure than custom platforms. The IT industry has developed various technologies for safeguarding servers that can be carried over to software-defined infrastructure to create robust and layered security. And since software-defined infrastructure is based on open platforms, industrial companies are free to work with any supplier they choose to adopt the latest technologies and process innovations.
Flexible industrial automation, powered by a software-defined infrastructure, will enable companies to react more quickly and economically to an ever-evolving market landscape.
Higher Level of Interoperability
In contrast to conventional single-vendor, proprietary automation solutions, open software-defined infrastructure platforms allow for a higher level of interoperability among COTS components from multiple vendors, giving industrial users more choice and flexibility. An open platform approach also makes it easier to upgrade systems and add features to keep pace with changing market demands—all at a lower initial and ongoing maintenance cost than proprietary systems.
Let’s take a closer look at how a software-defined infrastructure approach would work in an industrial automation application. The International Society of Automation’s ISA-95 model, a standard for integrating enterprise and production control systems, comprises four levels as illustrated in Figure 2. Levels 1-3 represent the operations and control functions, while Level 4 is the enterprise-level business planning and logistics layer. The premise of software-defined infrastructure in automation is that most of the functions found in Levels 1-3 can be run on COTS servers capable of satisfying the real-time performance requirements of industrial environments.
More specifically, as illustrated in Figure 3, software-based digital controllers, PLCs/DCS, SCADA software, HMI, process historians, and applications in L1-L3 can run in an industrial software-defined infrastructure. Servers interface to sensors, actuators, and other physical industrial devices via distributed control nodes.
In contrast to a typical IT data center installation, software-defined infrastructure makes the data center “industrial grade,” delivering the CapEx and OpEx benefits of an IT-based approach while satisfying such industrial requirements as high availability, real-time determinism, life cycle management, and hitless upgrades.
Compared to purpose-built proprietary solutions, software-defined infrastructure can deliver several cost-reducing benefits. It can lower hardware CapEx by substituting low-volume, custom computing platforms with a small set of high-volume COTS servers. These servers can be easier to manage than a sizable population of unique proprietary devices, lowering OpEx. Software-defined infrastructure would further lower CapEx and OpEx for logistics by significantly reducing the number of unique boxes that must be kept on hand for maintenance and the related costs to train and support staff on multiple unique articles.
A software-defined infrastructure approach has the potential to scale and expand with less effort because there are fewer wires, cables, and systems to deal with, minimizing connectivity related costs. Software-defined infrastructure solutions also take up less physical space near the industrial equipment they control.
By design, software-defined infrastructure -based systems require less field service support than traditional systems. In an industrial IoT environment, operators can monitor, diagnose and update software-defined infrastructure systems remotely and in real time, without deploying engineers, further reducing OpEx costs. If a failure occurs in the field, high-availability system failover mechanisms help reduce the need for an emergency truck roll.
Finally, having a software-defined infrastructure can mitigate the costs of avoiding system obsolescence. The decoupling of functions implemented in software from the underlying hardware and software platforms makes it easier to update systems over their expected service lifetimes.
When you combine the cost benefits with the added flexibility and the ease of keeping pace with technological innovation, the case for software-defined infrastructure as the next wave in industrial automation becomes quite compelling.
Now, let’s look at what is required in a software-defined infrastructure to realize those benefits.
Software-defined infrastructure automation solutions must run reliably and safely, gathering real-world industrial data and triggering real-time responses. To achieve this, a software-defined infrastructure must consolidate operations and control functions, and satisfy these criteria:
Low-latency virtualization: Software-defined infrastructure servers must support virtualization in order to run the diverse functions and applications found in industrial systems. The virtualization technology must have minimal overhead to realize real-time, deterministic performance for critical applications while optimizing resources for non-critical applications.
Deterministic networking: Fully deterministic, real-time communication via the IoT is essential for control functions in industrial environments. Time Sensitive Networking (TSN) achieves this by enabling a shared view of time and scheduling among industrial components.
High availability: In the event of software failure, software-defined infrastructure servers and applications must be able to perform automatic failover quickly enough to maintain control system integrity. Failover speeds need to be orders of magnitude faster than standard IT solutions. Carrier-grade telecommunication NFV solutions are approaching the automatic failover speeds needed for software-defined infrastructure. Virtualization technology facilitates failover in a number of ways—for example, restarting a clean backup software image without a reboot or turning control over to a full redundant server to avoid catastrophic failure.
Robust security: A software-defined infrastructure approach allows security technologies to be built in from the ground up across hardware platforms, middleware, applications, communications, and cloud infrastructure. The flexibility of software-defined infrastructure allows security solutions to adapt over time to respond to system and threat changes. Required technologies include secure boot, robust roots of trust (for example, Trusted Platform Module or TPM), digital random number generators, secure identities, local and remote attestation, anti-malware, data encryption, firewalls, authentication, authorization, and accounting (AAA), IDS/IPS, SIEM, and VPN tunneling.
Lifecycle management: Automation systems are typically expected to remain in continuous operation for years. Users must be able to perform lifecycle operations, such as software upgrades, live patching, capacity expansion, hardware updates and replacement, and physical and logical networking changes, without any loss of service. IoT solutions that allow easy installation, remote provisioning, and extensive monitoring of platforms, hardware, applications, and services are essential to maintaining system uptime and performance.
Enhanced platform awareness and monitoring: Software-defined infrastructure solutions need to support awareness of hardware and software status to guarantee required levels of service. IoT-powered platform awareness and monitoring capabilities enable automated resource allocation and reallocation as needed to adapt to change while maintaining performance, safety, and resiliency.
Best-in-class applications: Based on open x86 virtualization architecture using COTS hardware, software-defined infrastructure solutions must support the easy integration of IT technologies (Hadoop, Apache Storm, Java Analytics engines, Linux, and Linux containers). At the same time, solutions must implement operational technologies capable of satisfying real-time requirements (that are more stringent than IT) through the use of industrial strength real-time operating systems. System integrators and operators can then take advantage of the open platform to incorporate ISVs and best-in-class applications.
It all sounds fairly complex, but many of these infrastructure requirements have already been addressed by telecommunications networks that implement network function virtualization (NFV). I predict this will be a revolutionary year with respect to digital transformation in industrial automation and the IIoT.
Gareth Noyes is Chief Strategy Officer responsible for overseeing Wind River corporate strategy and mergers and acquisitions activities, as well as for leading the Chief Technology Office. Charged with developing the company’s long-term technology vision, he has Wind River positioned to address the evolving market landscape and disruptive forces such as the Internet of Things, the virtualization of the network, and the transition to a software-defined world.