Touring Tomorrow's Digital Factory

Inside the Trends Transforming the Industrial Space

60 min video  /  41 minute read View slides


Don Pearson

Chief Strategy Officer

Inductive Automation

Craig Resnick

Vice President

ARC Advisory Group

About this Webinar

Major shifts are underway in the world of manufacturing. A convergence of technology trends is driving demand for simplified, standardized, and connected system architectures that facilitate faster industrial-data analysis at an enterprise level. New opportunities abound — but how can industrial organizations transition from where they are now to where they need to be?

ARC Advisory Group Vice President Craig Resnick and Inductive Automation Chief Strategy Officer Don Pearson discuss a new type of IIoT architecture that can increase data throughput, provide greater agility, and improve enterprise-wide communication. Learn how IIoT could reshape the way industrial organizations implement system architectures, and deepen your knowledge of the key factors driving this movement.

Explore megatrends in manufacturing: 

  • Digital enterprise/IIoT platforms
  • Edge computing
  • Open enterprise architectures for the IIoT age
  • Virtual and augmented reality in factory environments
  • The factory workforce of the future
  • Cybersecurity needs and solutions
  • And more

Webinar Transcript

(The moderator, Inductive Automation Chief Strategy Officer Don Pearson, briefly introduces Inductive Automation and Ignition software, and then introduces the main presenter, ARC Advisory Group VP of Consulting Craig Resnick.)


Smart Production, Digital Enterprises & The Industrial Edge
Craig: Today I'm going to talk about the megatrends driving the digital factory of the future. And as Don mentioned, we certainly could not cover all of these trends in depth in 60 minutes, so what we'll do is, I'm going to scratch the surface on some of them and then look forward to some of your questions if we want to take a little bit of a deeper dive into some of these topics.

Craig: We will spend time on the whole digital enterprise, IIoT, advanced analytics, edge devices, we'll get into that. I'm going to briefly cover things such as 3D printing and additive manufacturing, get a little bit into robotics digit twin/machine learning. Talk a little bit about both virtual and augmented reality and their role in the manufacturing space. Briefly cover the workforce of the factory of the future, and how/if it's changing from the baby boomer over to the millennials, and what implications that has. Going to briefly cover the Open Process Automation initiative that Exxon Mobil is working on and ARC, we're kind of right smack in the middle of. Just touch the topic of cybersecurity and we talk about some of the issues going on, but again, cybersecurity by itself could be a separate webcast, just with all the topics and space in which that covers.

Craig: So as I go to the next slide, I think the one common denominator that you want to take away from this is that everything on the factory floor; both from assets, from mechanical products, electrical products, the tools that are used to perform the services, the people. Everything today must be connected. And I think that's really been one of what's the key drivers of Industry 4.0, IIoT, is just being able to make sure that all those devices are communicating with each other.

Craig: There's really no way for an enterprise today to survive, if they're not offering a completely connected enterprise from the lowest-level device on the factory floor, all the way through the enterprise, and being able to take advantage of not only the process information that comes from those devices, but also the business information that's used to help run the enterprise.

Craig: So, if we look at the whole concept of the digital enterprise, you're seeing that in many cases some of the production concepts of today kind of follow maybe the traditional ISA-95 model, as far as moving things up from the factory floor, up into the enterprise. But as we move over into the digital enterprise, that's giving you the response to this whole world of data-rich, the connected reality, being able to connect all those types of assets, and being able to leverage intelligence at the edge, and bring all that data to the cloud, while converting that data into actionable information by leveraging distributed and advanced analytics. Which is key for any of this digital enterprise to be able to survive going forward.

Craig: Many times, we look at situations where we get in to what we call, it's platforms. The dual digital enterprise really consists of these IIoT platforms, but it really ranges in scale. It can start from smart phones and tablets, like you might find an iOS platform or an Android platform on your phone or tablet. You may be moving up to some of the gateway and edge devices and some of those platforms; some that you might find from companies like an Intel or a Cisco. We're all certainly familiar with the cloud providers such as Amazon's AWS, or Microsoft's Azure, for cloud platforms, and then you move into some of the platforms of the industrial companies that make large rotating assets; such as you might find with GE with Predix, or Siemens with Mindsphere.

Craig: Or also, there's companies like Inductive Automation, that certainly have a very compelling digital enterprise IIoT platform, and I'm sure you'll hear a lot more about that a little bit later in the webcast.

Craig: So, if we get into the whole digital enterprise, it really creates opportunities to optimize digital industrial operations; let's say such as maintenance using APM, or Asset Performing Management solutions. It's also a tool that can help optimize production. Leveraging the fact that everything is connected.

Craig: The digital enterprise also creates opportunity to provide digital products and services because you're enabling smart products, you're leveraging an abundance of low-cost sensors, and embedded intelligence and analytics. And you have the ability to also provide after-market services by remaining connected to the assets through remote monitoring and diagnostics, preventive maintenance, hoping to prevent what we consider the biggest nemesis for all manufacturers and processes, and that is the nemesis of unscheduled down time.

Craig: Many times you'll hear the term of industrial edge, what is it, and what makes it different. And if, simply put, it's really the locating technology infrastructure located in or near production operations for collecting, and analyzing, and storing data. It can be leveraged in many applications such as analytics. It can be used in almost any industry, remote locations. One of the beauties of the edge, it does not require that IT skills be located at the site, because of this remote connectivity. What is driving it? It's really the criticality of having local data. Sometimes this data has unique security needs, which is why it needs to be, say at the edge. People can use it on devices even like PLCs or PACs or Drives. So, an edge device can also be a controller, for example. Sometimes it's very common in plants with a cloud, where people certainly have maybe some reservations about using the cloud, especially for real time application. And certainly companies like ARC that write almost 100 markets studies a year; edge devices are probably one of the fastest-growing areas in industrial automation as far as the rate of growth is concerned.


Advanced Analytics, 3D/Additive Manufacturing
Craig: Looking at some of these IIoT or Industry 4.0, your network edge; you're connecting these devices, and sensors, and assets. Sometimes it's leveraging gateway switches, routers, and wireless access points that can be used, depending on a company's particular cybersecurity policy regarding these gateway type devices. It gets data and analyzed information up into these enterprise applications that are often in the cloud. Whether it be ERP, or PLM, or asset management, or even analytics that might be based in the cloud.

Craig: The factory of the future is going to run on advanced analytics, and this is really going to enable things like machine learning; which will power operational intelligence to determine best practices and risk and optimize production operations. This is enabled by almost an infinite amount Big Data that manufacturers have in their possession, estimated by the Bureau of Labor's statistics that manufacturing actually has the largest repository of Big Data anywhere. The categories in analytic software that drive this is primarily prescriptive analytics, and we'll talk about that a little bit later. Bringing together Big Data, statistic sciences, rules-based logic, and machine learning to help find the origins of complex problems and then determine decision-based options to help resolve them.

Craig: We talk about some of the different categories of advanced analytics; and the most basic, descriptive and diagnostic, which really focuses on what happens, and why, past and current performance, important for reactive decisions in real time. Well what the problem is, that sometimes it really doesn't help make the proactive decisions and doesn't really help you head off problems that could occur down the road. And that's really where you see things like predictive analytics come in, which uses structured and unstructured data, machine learning, and various business rules, to kind of make predictions of future activity. And also as I just discussed briefly, the prescriptive analytics which really leverages knowledge bases to come up with options and implications, and that really helps with the automation of the whole decision processes itself.

Craig: If we look at some of the advanced tools that are used as part of advanced analytic, it's things like correlation/regression analysis, statistical analysis, SPC, using information. Again, we talked about having, if manufacturing has the most amount of data, being able to mine that data from typically, let's say, the process historian, which is usually the greatest repository of data in process plants; and putting this all together will help enable this artificial intelligence and machine learning. And also the neural networks which try to think, try to feel the way the human brain may think.

Craig: So, if we look at another talk trend here from 3D and additive manufacturing, and seeing that move into more mainstream. We're seeing it in aerospace and defense, automotive, certainly the machine OEMs and some of the oil and gas industries; mostly on like the big rotating assets like pumps and turbines. This being adopted usually for the additive manufacturing or production parts, but really what its niche is, is it can produce sometimes these complex parts that are different to manufacture using some conventional techniques, even conventional CNC machines. So it can create these very, very, complex geometries and that give the manufacturer sometimes the certain tolerances and the surface finishes that are specified. One of the things it also does, of course, is it enables, rather than having to tie up cash in inventory, it enables you to create the products as they're needed, and timing it as such. So it's certainly good for the bottom line as well.


Advanced Robotics & Machine Learning
Craig: Next thing I want to talk a little bit about is advanced robotics. Sometimes we refer to them as “cobots” that are moving beyond the production work cells, working with humans and other robots. And we're looking at this almost as the worker's assistant, because it's enabling production lines to kind of leverage these physical cyber systems and getting to the point where these systems can self-optimize and self-heal and be able to run autonomously. Again, this wouldn't be possible if you didn't have the ability of the connectivity, IIoT, edge devices, advanced analytics, and also the ability to create what we refer to as the digital twin.

Craig: And a digital twin would be, picture a software simulation of any mechanical or rotating asset; whether it be a pump, or a turbine, for example. It simulates every aspect of the physical product, can be used for testing, simulating, commissioning the digital twin even before the product is actually put into commission in the real world, and really adds, from a manufacturing perspective, adds a tremendous service that they can provide the users of those assets.

Craig: We've also talked a lot about machine learning, and what machine learning is, is it's really taking Big Data, doing a lot of data mining, applying advanced statistics, and leveraging a lot of the advanced analytic algorithms to create this pattern recognition. Which is, again, artificial intelligence that when put into machines really enables machines to mimic human behavior. And we talk about machine learning, one of the great aspects of that is that it really helps to teach computers to make decisions without having to program the computers, because what you're really doing is, you're letting history, an abundant amount of historical data put into the necessary analytic algorithms, and you're actually programming the computer based on these terabytes and terabytes of history that you've been able to accumulate. So with that, I'll turn it over to Don.


Industrial Internet of Things
Don: Great, thanks so much. I know you've covered a tremendous amount of territory over the last 15 minutes, and even as you said at the outset, and some of those topics could be uphold all on themselves, without a doubt.

Don: Just couple of these bullet points I wanted to maybe make a couple comments on, and have a little bit further discussion on. There's been a tremendous amount of hype around IIoT, and the need for Industry 4.0 and IIoT to progress, but I sometimes question the amount of real progress organizations are making when they're demanded to keep plants running 24 by 7. How do they make transformations? How do they break with the old model and the new? We totally agree with your point that achieving this digital transformation is going to be critical, but I maybe want your comments on this sort of gap between thought and action in this area.

Don: The comment we had from the World Economic Forum that 84% of business leaders expect IIoT to destruct their offering models within the next five years. I think that's a very powerful number. Most everybody knows it's going to disrupt it. But only 7% of those leaders from this study have a comprehensive IoT strategy, and 73% admit to having none at all. So, when I see that, I see a gap between the need, the requirements, and the actual action. Do you think organizations are actually starting to take IoT seriously and really develop and strategies, and take action a little more rapidly?

Craig: Well, Don, they're definitely taking it seriously. I think the reason for the disparity between the high percentage of interest and the very low percentage of actually doing it, is really the fact that there's a lot of confusion in the marketplace, and everybody is so risk-adverse that they're afraid to go on the wrong path and make the wrong decision. So, they stay, and it's almost like they're in information paralysis. They continue to gather, and gather, and gather, but to get anybody to sign the purchase order, or make the decision for fear that they could be doing something wrong, maybe going in the wrong direction, I think has been offsetting the ability to move forward.

Craig: So one of the things that it's the responsibility of companies like Inductive Automation and analyst firms such as ARC, is to really help with the education process to make sure that we help these companies feel comfortable that they are making the right decision. And we always advise customers to go with what we say is a low-hanging fruit strategy. Where, let's go after the problem areas, and what are the problem areas? Areas that are causing these companies to have unscheduled downtime. And let's work on those areas first, rather than trying to come up with some enterprise-wide, sweeping IoT or Industry 4.0 strategy, ripping and replacing out a lot of equipment.

Craig: We find that, let's help people walk before they can run, and that tends to kind of help ... well, hopefully that will get that 7% figure growing much faster.

Don: Yeah, I think you're onto something there. I think we really need to work as organizations to help in the migration of peoples’ thought process as well as their operational process in moving forward.

Don: Another comment I wanted to make ... you had a lot of data on that slide about all the different kinds of platforms, and I think we agree very strongly that platforms are the direction to go. I mean, certainly Inductive Automation, with the Ignition software product, that is the platform in the industrial application space, and I know one of our strategic partners, Arlen Nipper, co-inventor of MQTT, actually said that we need to start a shift from coding on operating systems as a mindset, to building tools and applications on top of platforms. It's a different perspective, and if we provide tools and structure for integrators and end users to build on, then they're going to be able to build on those platforms and they'll be fundamentally a lot more maintainable if they're building applications and schedules on top of platforms. But how important, from your viewpoint and maybe ARC's, is this move towards open, interoperable and maintainable systems?

Craig: Well, let me say this. You cannot achieve universal connectivity of, whether it be Industry 4.0, IIoT, or any of even the certain various country initiatives, unless you adhere to open standards. Proprietary open is not going to cut it any more. You cannot have standards which are really good for certain categories or certain groups of companies. You really need to be at a point where ... And certainly MQTT, OPC, are certainly clear examples of communication standards that have been widely deployed, and I really think that for this to be successful, everybody is going to have to take a complete, open perspective, and I will cover that a little later when we even talk about the Open Process Automation initiative, which will certainly help accelerate that as well.

Don: That sounds good. You're right on in that area. You can't have an OT/IT convergence and have Big Data be of any value, if you don't connect to the edge.

Don: I've heard numbers upwards of 90% of data right now is stranded in the field on those edge devices, and it's not getting in to the network where it can be of value into the enterprise. So, certainly we support that. We have a series of models for edge computing, panel enterprise, and MQTT; and clearly see that bringing edge devices into the enterprise connectivity is critical. So I just, maybe a comment, I appreciate your comments on that.

Craig: Sure, Don. I think one of the things that's going to be critical with edge is that there's always going to be a need in the industrial market, there's always going to be a need to have local data. Even as you've had greater and greater acceptance of things like the cloud, but there's always going to be that feeling that if the data is processed on site, at the device level, that's as real time as it gets, because this is always going to be an industry that lives on real time. And even though, as with a lot of the cloud applications and business applications, we're getting far more real time from the business side as well. But to get at that microsecond, millisecond level of real time that's needed in manufacturing, I think that's always going to be the driving force for having a lot of that intelligence and the ability to do that edge analytics right at the device level. So, I think that's really one of the things that's going to drive it as well.

Virtual & Augmented Reality
Craig: We hear a lot about virtual and augmented reality. And some of the things that ... what are some of the differences between the two and what are some of the similarities? When you get in to augmented reality, you're in the situation where you are now looking at the asset and you're seeing the real part of the asset, but you're also having information that's supplemented about the asset. Whether it be, for example, based on your GPS location, or based on maybe your ability to scan a QR code at the asset. Maybe now you can at the device and have the drawing brought up for you. Maybe as you're looking at the device, you're having HMI information that you'd get from a product like in Ignition, that would give you some of the parameters of the product. So augmented reality, sometimes people use the term mixed reality, allows you to kind of have that combination. It's really going to be, I would say, would be the best friend of the maintenance person, because it's primarily going to be viewed on a headset type of device, a HoloLens or a similar type of product. But, we're very, very ... augmented reality is really going to be the maintenance tool of the future.

Craig: What virtual reality does though, is it really enables you to simulate almost any situation that could happen on the factory floor, because now you're complete immersed in that environment, and there's no other external ... anything else external that's part of the experience. And that's certainly what you're seeing in a lot of gaming. It's also technology they've been using for years with pilots, for when they go to the simulator to simulate things that we certainly don't ever hope happen in the real world, but at least trains the pilots to be prepared for that. And we do the same in plants for preparing people for potential fires and explosions, and help to create disaster planning scenarios.

Craig: So, both of these technologies really offer some tremendous potential in the plants and we're starting to see it being used. I cannot emphasize enough how virtual reality is really going to be the training of the future, and it's really going to be the situation where no longer can they say that we want to make sure that we can possibly create any disaster-oriented scenario and show people, how should the operators, and the maintenance people, and all people within the plant, how should they respond to that. And it's also something that again can be, the training can be done in an offsite, in a portable fashion, again just as long as the person has the necessary headset type of device to do that.

Craig: And augmented reality, again using things like DAQRI Smart Helmets, there's a company called RealWear, Microsoft HoloLens, I think that the key here, with augmented reality, is sometimes people have said, well can't we get that if we're walking around the plant with our smart phone or tablet and be able to kind of view some of this information on that? But we find with augmented reality is you really need to be immersed in the situation. So as the maintenance technician is looking at the product, the information they need is right at the eye level, and they don't really have to take their eye off the product to be looking down at another device. So, we feel very, very strongly that again, it's really going to be hard to deploy any type of service organization without having augmented reality tools as part of the whole experience.

Craig: We look at the workforce of the future, as we start to see the baby boomers retire and being replaced by millennials. You now have people who are very, very, well versed, and skilled, at know the plants and the processes with a database that's often in their head, and now you're moving to people who are going to have very, very, strong skills at using a lot of the tools, but maybe don't have quite that experience that they can't replicate from 30 years. So that's going to really change from doing the manufacturing, to actually monitoring the automated processes. So it's really going to change the role of what the plant worker of the future is really going to do.

Craig: It also means that finding talent is going to be an issue. And actually one of the strengths of leveraging a lot of these new technologies, like augmented reality, virtual reality, analytics, is it finds an attractive place for many people that are graduating with engineering and computer science degrees, if they recognize they can use that technology in the workforce. And sometimes we all need to do a better job of conveying that plants are not just 25-, 30-, 35-year-old equipment that leveraging technology of the prior generation. That is as these technologies become implemented in the plants, you're going to see far more opportunity for attractive positions for the millennials. It also, of course, requires that this create a collaboration between industry and government to make sure that the technical schools, for example, are producing graduates that have these skills to use these tools, so they can step right in to some of these new jobs, leveraging these technologies.

Craig: And interestingly enough, at one time, you would never have seen a job description for a data scientist in the factory floor. But now that is one of the fastest growing positions in manufacturing and in all of industry. Being able to contextualize and analyze data, visualize the data, come up with the right intelligence. So as important as engineers and operators and process experts will still be to the factory of the future, but they're going to be standing side by side with these data scientists. And where people find how valuable the whole commodity of data is; sometimes people use the expression that data may just be “the new oil.”


Open Process Automation
Craig: I do have one slide on the whole Open Process Automation initiative that was very much driven by Exxon Mobil. As you look at this chart, this is kind of a schematic of an envisioned architecture. The new components are colored yellow, existing systems would be in light blue, and new capabilities are going to be broken up into three major areas. There's a new operation platform, as we all talk about IT and OT convergence, but this new type of OT platform that will be implemented is really going to be really highly standardized IT like software and hardware. In our view, this likely will be implemented as using on-premise cloud platform, maybe some additional real time capabilities. It's really going to make extensive use of virtualization and open-source software.

Craig: Second is the real-time data service bus, and it's really a set of data services that tie the system together, and enables incremental expansion and change, and this may also be implemented using open source software. But regardless of the definition of the services, it'll certainly be public, and certainly standardized.

Craig: Third is a dedicated, single-loop control, and we use an acronym DCN, which would be called a distributed control node. It's a very highly distributed edge module. It should be a great many of them in each system, and in many cases, the DCN may regulate just a single control loop. This has advantages of lengthening the span of the automation control and being able to do a better job of predicting where there would be a failure, for example, and be able to address that failure quickly. And today's DCS controllers manage hundreds of loops, rather than one, and therefore much more critical components, but over time we’re seeing these DCS functions migrate to either DCNs or goes into the future.

Craig: One of the things that's driving this whole phenomenon for open process automation is people ... Exxon Mobil does not want to be tied to any specific system. They want to be able to have a system that's dynamic, rather than static, and not have to make an investment in buying a system they're going to have to amortize over a 30-year period, for example. So they want to be able to invest in new systems to become the DCS of the future, that can leverage the latest in technologies an open standards, and based on the interest level in this from all of the end users that are joining the Open Process Automation organization, the suppliers, such as Inductive Automation that are involved in this group; ARC is very optimistic in the future of open process automation. So with that, let me turn it over to Don.

Don: Craig, thanks so much. I definitely appreciate your comments. There is a new generation of engineers. They're very familiar with IT, web based, open sourced type technologies. And I think we as suppliers, and those advising people in the industry like ARC, need to help this OT/IT convergence and things that meet the needs of OT and also are IT-friendly, sort of help break down that wall, and I think are going to be very conducive for that next generation to take up with a little bit.

Don: And on your subject of data, boy, is there a lot of data. I constantly hear the comment on the 5 V’s of data, but certainly the volume, the velocity, the voracity, and the variety of data coming at people. You need these data scientists there in order to get at the fifth V, which is value, for God’s sake that everybody's trying to get at.

Don: You mentioned open process automation, we're very involved in the initiative, I'm supportive of it. Is there anything you'd think that is additional to how people are getting value out of data?

Craig: Well, I think the idea of the getting value, is first of all making sure that all that data is what we always refer to as, the single version of the truth. Meaning that the data is going to be true regardless of where it comes from and regardless of how it's been analyzed and processed. And two, to really make sure that data is utilized at all points in the enterprise. Again, not only certainly from the manufacturing and automation side, but to make sure that that data has equal value in the production management and ERP systems, and the business part of the enterprise; to really make sure that that data is truly being maximized. That's really how you're going make sure you get the appropriate value.

Don: Thanks, Craig.


Craig: Now, I'm just going to briefly get into the whole area of cybersecurity, and granted, we are, again as I said earlier, this in itself is an entire webcast. But when we talk about some of our customers, as far as what are they trying to protect. And from a plant prospective, they're always talking to us about endpoints, it should be devices that are used; remote SCADA systems, anything that has a direct impact on operations. Whether it be servers, work stations, PACs, PLCs, DCSs, RTUs, embedded systems, and maybe any PCs, laptops, that are brought into the plants in any sort of remote HMI/SCADA for maintenance. They're really not as concerned with the people’s personal devices because they feel as though many times they can firewall those devices, or companies that maybe don't have a BYOD, or bring your own device, program within the plant.

Craig: So when they think about endpoints, that's what they're thinking of, and when they think about networks, they're certainly looking at their demilitarized zones, their networks for connecting Level 1, 2, and 3 devices. There are traditional networks that are used today, to make sure that all the devices are connected both to each other, and certainly getting up into the MES and production management. But they're really not thinking in terms right now of the networks by applications outside the plant. So again, this is also very, very, inwardly focused when it comes to looking at it from a cybersecurity protection perspective.

Craig: And again, these are the info’s that are really based on some ARC survey data, and when they talk about, for example, what's actually needed to secure these systems; when they talk about on premises, it's really network and endpoint security solutions. The things that have become commonplace such as application whitelisting, deep packet inspection, firewalls. But it's really a matter of making sure that anything that's new that's being installed in the plant is designed with cybersecurity in mind. However there's no illusions that they recognize these intrusions are going to continue and are going to remain a serious concern for these companies.

Craig: Some of questions that they always have for us is, is there such a thing as cybersecurity maintenance? And how major of an issue is that? Have we kind of reached a practical limit based on technology, or is it just a matter of, it's a journey not destination? Which we're trying to see if we can stay one step ahead of it. Are there other solutions that can address some of these gaps in the current solution set that may be coming from the IT world? And this is certainly going to be helped with IT/OT convergence. Do we just take a different mindset and assume the hackers are going to get in, and shift the detection to rapid detection, so we can hopefully find out about it faster than ... sometimes we always hear these stories about people that hear of the fact they were hacked on months ago, and that the damage is just starting to occur, and they really have a difficulty trying to respond. So, again, that's just some of the questions that we're having. So, let me turn it over to Don right now.

Don: Well, Craig, I know we're just touching on this, right, and I want to give you a change to move on because we've done some webinars on cybersecurity, we're certainly strong advocates of the fact that it's not just technical problems obviously, it's a mindset shift that needs to take place. You called it, it's not a destination it's a journey, I think that's very critical. Because when you think about it, we're pleased to work with companies like Bedrock Automation, because they clearly feel that intrinsic cybersecurity needs to be built at the control level, and we fundamentally agree with that philosophy. Security measures are much more effective built into an architecture and to the technology rather than bolted on as an afterthought. It's got to be part of everything, as you've covered I think, very well.

Don: Is there any just quick thought you may have before we move on, on just the mindset shift that needs to take place to vis-à-vis security?

Craig: I think the mindset shift is that ... the excuse of saying, ‘Well, my only way to be secure is to not be connected,’ is not going to fly. So, you really have to ensure that you're just making sure that you are connected and just leveraging all the necessary precautions, and working with the cybersecurity experts, again many of them coming from the IT space. So, it's really a matter that you can't use that as an excuse. And the other excuse you can't use today is to say, I'm just going to actually stay with some of my older technologies, or some of my older operating systems because those are not really the target of many attackers; when in reality a lot of those older product and older operating systems, where they're no longer doing update and patches, and no longer have the support, are actually some of your most vulnerable areas. So in reality, that should give you an excuse to upgrade, rather than an excuse not to upgrade.

Craig: So I'd say those are just a couple of points that I would make to those that are maybe a little bit cynical about connecting and maybe moving things up into more updated product life cycles.


Open Enterprise Architectures in the IIoT Age
Craig: So, I think for our final section here, we're going get into talking about the whole concept of enterprise architectures, and linking the factory floor operations, or the business operations, kind of across the entire corporate entity.

Craig: It's certainly not new, it's been around awhile and in most industrial sectors. Making the concept of reality continues to be a challenge. It's certainly going to be true for companies that certainly they don't have huge IT staffs or huge budgets. The whole IIoT, and it's kind of the promise of accessing, aggregating, and analyzing data from previously stranded assets and systems. It kind of represents, in some case, a further disruption, because again, how do you actually go ahead and execute that? You have Industry 4.0, we have IIoT, and it's certainly, it's pretty hard to go anywhere without reading about one or both. But the reality is, most industrial organizations have different data, and hardware, and software, at each plant. They don't really have a smooth way to get plant data to the corporate level. I think especially when you see some of those statistics that Don cited earlier as far as those that are thinking about IIoT, versus those that are actually executing it.

Craig: But when this has spread, the demand has grown at the corporate level to get more of this data from plant operations, putting it in an actionable information, they see the potential of how these companies can save millions, tens of millions, hundreds of millions of dollars sometimes. Better job decision making, enabling centralized management, paving the way for new technologies like we talked about with predictive analytics and machine learning. But successfully transitioning to the enterprise system, it's no small task, and I think that's the reason why we've had some of the resistance to change, based on the numbers again that Don cited.

Craig: So, to kind of meet these new demands, we need automated processes to deliver this plant information to the corporate level. It has to be accurate, standardized, efficient, and secure, and as I said earlier, it really has to make sure it's always going to be the single version of the truth. You know, for most organizations, a first step towards effective enterprise architecture would be shifting away from the separate buckets of IT and OT, and going through this whole IT/OT convergence that ... and I can tell you from the ARC Forum, the interest in IT/OT convergence is pretty universal, based on all our client companies.

Craig: The challenge of building an enterprise architecture, it really can't be addressed with top down thinking. And it really must be built from the bottom up, right from the sensors, and always thinking in terms of what, how’s that information going to help achieve the objectives of the business. You can't think of each plant as an island, you really must look at it as kind of a larger corporate system with common standards and common data transport mechanisms, and the companies have to ask themselves, what do they need to do at the plant level just to pull up the enterprise. It requires secure connectivity. So, the idea is, what can we do to make everything look like a universal enterprise and stop thinking of us as not manufacturing versus operations versus IT, but looking at this as one converged company.

Craig: The organizations really need to be balancing the whole need for security, and the need for data, as we talked about earlier. Existing technologies can be used to get data, a central system in an open format, without compromising operations. And I know there's air gapping was often deployed in the past as a security measure, but this world kind of isolates OT from IT, and obviously we're trying to converge that. So it's really not going to serve any enterprise needs going forward.

Craig: So, really, to ensure proper security, data really needs to be encrypted, and shared across sites, for the plants that are using things like PLCs and PACs. It's often if you can get an edge gateway next to the PLC or PAC, and get that data into an open format, it's also going to help it be more secure.

Craig: To build a system with a centralized visualization & administration, it's really pretty essential to start developing standards across the entire enterprise. Bringing different plants, from PLCs, PACs, tags, address schemes; to put that in some sort of a standardized system, you really need to employ an open-source protocol that's supported by a multitude of applications. Whether you're talking about OPC UA, whether you're talking about MQTT to get data from devices, or SQL for working with SQL databases, or various APIs such as OPC UA, SOAP, or REST to help with the integration of other systems.

Craig: And once the organization chooses an open standard, they can standardize data models, so the data from all the plants looks the same when it's sent up to corporate. And this is a much more practical approach than gathering different looking data from different sites and then trying to go through a translation process at the top.

Craig: The protocol that's chosen the model must have, or should have, built-in encryption, have stateful awareness so it tells you when whether or not you're connected to a specific device. The combination of encryption and stateful awareness kind of puts one approach for getting operational data to the business side in a standardized way, and doing that without compromising operational security. OPC UA and MQTT protocols can offer that built-in security and give you that stateful awareness that's needed.

Craig: Another question is, how you enable the enterprise to get the data it needs without stifling innovation or interrupting operations in plants, which is obviously unacceptable in a world where unscheduled downtime is completely unacceptable. So it's open architectures that a couple of the applications from devices, is a possible solution to that. So, if you can see on the slide, there's an example of a coupled architecture using a poll-response protocol is on the left, a decoupled architecture using a published subscribed protocol is on the right. In the conventional architecture, intelligent devices such as PLCs and PACs are coupled to applications with proprietary protocols; and any application can interact with any connected devices, and typically in these architectures, the HMI/SCADA software communicates with the PLCs and while it's not its intended purpose, sometimes the software’s often used as middleware, because it has the protocol needed to do so.

Craig: But when you're in a decoupled architecture, the applications are not connected to the devices, but the devices are connected to the infrastructure, so the applications can subscribe to the data that they need. So rather using HMI/SCADA software as a middleware, these decoupled architectures can use some type of message-oriented middleware, and that's where MQTT comes in. And as you can see from the example, devices publish data by exception to a central MQTT broker, that can be done on premise or in the cloud, the HMI/SCADA software can subscribe to the data, and other applications like ERP, or MES can also access that data; so sometimes we look at is as a data buffet in which various systems and tools can take advantage of just the data they want, and instead of having integrated programs with each other, the programs have direct access to the data. So it also offers plug-and-play interoperability anytime any new devices or sensors get added.

Craig: So, by making the device the single source of truth for the tag information, these decoupled architectures can end up saving you many, many man-hours, and really allow anyone in the enterprise to have the same data to base their business decisions on, so again, emphasizing that whole single version of the truth. So in short, the type of decoupled architecture can provide the functionality needed to realize many of these benefits of IIoT and Industry 4.0's, better and simplifier connectivity between sensors and applications across the enterprise.

Craig: I think we see that MQTT is gaining a lot of traction as an IIoT messaging protocol. Where it's really differentiated is its kind of lightweight overhead, a two-byte header, its publish-subscribe model, bi-directional capabilities so it really requires only minimal bandwidth. The MQTT collects data from a multitude of devices, can transport that data to the IT infrastructure, can be used in real time, mission critical HMI/SCADA systems, and it's unique in the fact that the payload is really data-agnostic.

Craig: So it gets used in a lot of applications outside of the traditional industrial space, like Facebook Messenger, Amazon AWS IoT service, IBM's messaging middleware systems. Ignition's OPC UA server also provides connectivity to multiple protocol and its open API facilitates interaction and data sharing between applications, and also as well as the development of protocols including MQTT.

Craig: So, another important reason for using the open standards in architecture is really the vast number of companies that have these multiple disparate brownfields. To integrate a brownfield facility in a real enterprise solution, you're really going to have to standardize the data and models, and realistically this will require some financial investments to implement tools at the plant that can convert data to a known interoperable format.

Craig: As you mentioned earlier, edge gateways can be installed in PLCs in the plants, and poll PLC data and get it into kind of a decoupled message-oriented middleware infrastructure. It can be decoupled and parallel with existing HMI/SCADA systems that are directly communicating with the PLCs, and then eventually the organization can start transitioning everything over to the new architecture.

Craig: But this requires investing in some new hardware, but it's an investment that certainly is going to offer excellent ROI potential. It's also important to acknowledge that no one solution can do everything the organization needs. But it's vital to have a solution that can integrate with tools for business intelligence, and machine learning, and open source software, IIoT, edge computing, business management, and a lot of the subjects that we've been discussing today. So, think about whether you can integrate the solutions and architecture into the plant now using these tools, but going forward invest in solutions that can integrate with those solutions, otherwise you're going to create your own island, and we certainly don't want that.

Craig: So, let me turn it over now to Don, has some final discussion.

Don: Craig, thanks so much for doing that. I think I want to mention just one thing actually, as we move forward. I know you covered this very well, but we certainly feel that this new architecture is critical in everything we're doing at Inductive Automation is trying to support the evolution. So, industrial organizations, and that brownfield challenge, can absolutely take advantage of that challenge. I'm also going to comment that, to all of our attendees, in the handout section on your console, if you click that handout section, there is a white paper that Craig and the team at ARC put together called “Creating Modern Open Enterprise Architectures in the IIoT Age.” It was last September that it came out in the ARC View white paper, and anybody that wants a full copy of it from that last section that Craig dug into pretty deeply, but you get a full view of it if you just go to the handouts section, you can get your own copy, and read it and see how it applies to the organization you're involved with.

Don: With that Craig, let me let you wrap up real quickly, and then we'll see if we got time for a couple of questions here, since we have some questions in the queue.

Craig: Okay. Well, I think to wrap it all up, I think the factory of the future really is now. It's here today, and it's really just how we're applying these technologies; whether it be artificial intelligence, machine learning, augmented reality, virtual reality, prescriptive predictive analytics, edge. And I think the thing is, instead of trying to spend a lot of time trying to become, always educating yourself, but not deploying. I can't emphasize enough is to let's look at the areas of vulnerability in your plants, where you're having the most problems, where you're seeing some loss of production, where some of your KPI's are falling; and rather than trying to do the entire plant at once, let's try to find ways of trying to implement some of these technologies in the problem areas. And I think once you start to see the return on investment, and the benefits of the technologies, I think then all of a sudden the 7% will become 70% in a very short period of time.

Craig: So with that, let me thank you Don very much, and looking forward to some questions.

(Craig and Don answer questions from the audience for the remainder of the webinar; please go to the webinar recording to hear the Q&A.)

Posted on March 1, 2018