Home Digital Transformation How to Effectively Implement Edge Computing

How to Effectively Implement Edge Computing

by DoShik Wood

Automation World’s Director of Content/Editor in Chief, David Greenfield, recently sat down with Aric Prost, Senior Director of Business Development, at Stratus and John Vicente, CTO, to discuss the state of Edge Computing deployments, considerations for end users, and the path ahead for Edge Computing in the digital transformation journey. To hear their insight, check out the podcast below.

Where to start (2:03)
Industrial PCs vs. Edge Computing with virtualization (3:27)
Deploy Edge Computing with System Integrators or in house? (4:40)
Operational benefits of Edge Computing (7:17)
Benefits for manufacturers (8:08)
The deployment process (9:29)
Edge Computing vs. traditional automation and control (11:37)
Edge Computing for predictive maintenance (16:36)
High availability vs. fault tolerance (18:09)
Benefit of Edge Computing for OEMs (19:52)
Remote operation and security (22:25)
Latest trends for manufacturers and Edge Computing (24:53)

David Greenfield, Director of Content/Editor in Chief, Automation World: Welcome to the Automation World “Get Your Questions Answered Podcast” where we connect with industry experts to get the answers you need about industrial automation technologies. I’m David Greenfield, Director of content for Automation World. The question we’ll be answering in this episode is how do you effectively implement Edge Computing.

Now, Edge Computing is a topic that’s been receiving a lot of attention for the past several years now for a variety of reasons because it brings high levels of data storage and analytics on site, provides security for remote access to equipment, reduces downtime through better equipment insights, and even provides the ability to deploy virtual machines to reduce on-site computing resources and enable operational backup. So as so many capabilities, it’s not a surprise that there are a lot of end user and OEM questions about Edge Computing technologies. And that’s why I’m joined today by Aric Prost, Senior Director of OEM and VAR programs, and John Vincente, Chief Technology Officer at Stratus Technologies. Now, since so much information is available, describing what Edge Computing is – and you can access plenty of it on the automationworld.com and stratus.com websites. In this episode, we want to focus on the actual implementation of Edge Computing. So with that, let’s start off with some basics.

David: Eric, you know, where should manufacturers and processing companies start when thinking about their potential adoption of Edge Computing technologies? And by that, I mean, how should they review their equipment and operations to determine where Edge Computing can provide a benefit?

Aric Prost, Senior Director, Business Development, Stratus: As with everything, probably developing a strategy is a great place to start here. You can implement these strategies in a number of different ways. [Leaders] can think about [Edge] Computing as part of a large project that they want to do for digital transformation, or something that can be done in smaller phases, and determine which way they want to go from that perspective. They should definitely do a cost/benefit analysis and look at the ROI. They can think about that in terms of what applications they want to run, what goals they want to achieve. They can pick applications like thin clients, and run HMI software and control software, historians analytics, cybersecurity, the applications are nearly infinite, for what they can do on Edge Computing. And so as they think about it in terms of what they want to accomplish, then they should think about which areas are the best to go after first, if they’re going to do it in phases, going after the more high-value areas, faster return areas, and looking at industrial PC clusters that can be replaced with virtualization on an edge computer are great places to start.

David: So to explain the process of replacing industrial PC clusters with virtualization on an edge computer – how is this done? And how does it work in practice?

John Vicente, CTO, Stratus: Sure, Dave, the formal usage or category of activity that Aric was talking about is called workload consolidation. And in simple terms, it’s consolidating a larger number of PC or physical assets, performing, let’s say, a single function or an application. And you’re consolidating that to a smaller number of advanced virtualized edge computers or servers. It enables both a reduction of operating costs as well as well as capital expenditure costs, due to what is called physical sprawl, essentially requiring greater resources to manage those assets and their lifecycle. The concept isn’t really new. You’ve seen it in the enterprise cloud and telecommunications industry, it is now just taking a greater hold in terms of opportunities within the Industrial IoT space.

David: So Aric, is this process of reviewing equipment and operations, something that outsource outside organizations like system integrators or consultants are needed for or is this something that can be done pretty easily with in house operators and engineers?

Aric: That varies, it depends on the capabilities of their current staff, their overall digital transformation strategy, and the equipment they have to some degree. Different edge computers are developed based on that particular manufacturer’s experience, their experience with target market. It’s their understanding of what their particular customers are asking for. In a manufacturing environment, something that was developed designed with an OT professional in mind can be a big advantage.

Because those environments don’t have some of the IT-type resources that you might find in different verticals or industries. There’s a number of really good solution builders and systems integrators out there that can assist in developing their initial strategy for a customer, as well as the initial design after that strategy.

And [when] the goals are determined they can help them with designing the solution and do that in terms of what the current architecture inside the plant is. System integrators are really good at recognizing potential pitfalls and strengths within an architecture and then shaping a solution around that. And so their experience working in these industrial environments day in and day out, can certainly be something that’s leveraged the advantage of the customer.

David: So, John, would you say that for most manufacturers, and then by that, I mean those without large amounts of Edge Computing expertise already in house? Would you say it is advisable for them to work with an integrator to be able to get the most benefit?

John: I would for most manufacturers, but really, as Aric was talking about earlier, they have to look at this as part of their broader digital transformation initiatives. And so be careful in terms of what specific skillsets in fact they want to relegate to solution builders or integrators. But at the same time, this is a journey for most of these companies. And it’s a question of what expertise, do they really want to bring in house as part of their digital transformation initiative and develop those skills and that expertise within their own workforce. So it’s a combination of both but again, really looking at this as a broader initiative – that’s the key point.

David: Understood, thanks for explaining that. No, at this early review and assessment stage that we’ve been talking about here so far, Aric, can any specific operational benefits be reliably determined or projected yet?

Aric: Yeah, there are in-house tools that many customers have that they use to determine ROI and make decisions on what technology they will deploy. And there are generalities, a typical customer sees this much of an advantage or gain by employing certain applications. And then our company, Stratus, we’ve developed some tools as well, that can help with that type of a projection and positioning using those generalities that we see within an industry and within the customers that we’ve served. Giving a just a general kind of view of here’s your ROI, here’s how quickly you can expect to achieve achieve it. And these are the areas that we believe that it comes from.

David: Can you give me an example of how those tools have helped some of your manufacturing or process industry customers?

Aric: Sure, we have customers that range from marine applications to pharmaceutical, metal forming, and Oil and Gas. So you’ve got a number of different types there from discrete to process-type customers. They’ve leveraged Edge Computing to decrease maintenance costs, utilizing maybe machine learning, or predictive maintenance that might be built into the hardware, if it’s designed for manufacturing environment. They do workload consolidation to John’s earlier discussion, one of the previous questions, and they’ve taken advantage of on machine architecture to reduce latency, and have really leveraged the OT-friendliness of our platform to realize deployment time savings to the degree that they say for each deployment can save eight hours, which is the equivalent of a full day installing those because of the OT friendly nature of our hardware. So you think about that, that’s not just cost savings in time that you don’t have to spend out there doing that commissioning and deployment. It’s increased production of that end customer have potentially a day or more as well.

David: Once a user has identified an area or areas in which Edge Computing could provide a benefit to them, what are the next steps in the deployment process?

Aric: The quick answer there is, let’s determine if they can do an in house with the skillsets that they have already or if they’re going to need partners. As part of that strategy, many leading manufacturers have developed even specifications for equipment that allow continuous integrations of any integration of any new assets. They buy into their current digital environment. And so those specifications can then be passed out to any vendors that are providing equipment that goes into the plant, and can be understood by any integrators or firms that are helping work to further develop their strategy for digital transformation.

David: Can you share some insights about the specifications you were just mentioning and how smaller manufacturers in particular can potentially develop this for themselves as part of their edge implementation strategy?

Aric: Let’s say they pick like one key machine or one area and in the plants, and they want to improve the UI, or that’s, you know, a typical goal that we’ll see out there, they could do something like, determine what applications they want to do to accomplish that, and select maybe a machine learning partner, doing research on the web, talking about a number of different companies with Stratus. We have a few partners that we work with that we can suggest and kind of get the wheels turning in for whatever area there might be. And so we’d be happy to have those conversations, as they determine how they want to do it. And then determine, “Hey, can we do it in house,” as we discussed before or do we need a partner and then selecting an integrator partner that’s got experience in that space, and then [selecting a partner] familiar with their particular operation is a great next step.

David: John, would you say that Edge Computing is changing traditional automation and control architectures in any certain ways?

John: Sure, If you look at Edge Computing, the term, it’s a broad term. But what I would consider the high-value usages that we’re seeing in today’s transitional environment is, based on what we’ve talked about earlier, workload consolidation, leveraging virtualization technologies, containerization technologies, it’s really not only helping to address the costs, but also offering new and innovative ways to approach the automation and control environment. And many of these we talked about, and have been talking about from the IT or cloud industry. And there’s a number of emerging providers that are coming in with solutions like edge analytics, or AI, cybersecurity, advancements in service management, including fault tolerance, high availability, which, you know, we’re, we’re well known for.

There are a broad range of applications, and system management capabilities, that have evolved and matured quite well in the IT industry are starting to come in and creep into the automation control environment in ways we haven’t really seen. And the enterprise cloud or telecom industries.

In the longer term, it’s inevitable that control environment, the last mile, where there’s more deterministic or safety requirements, much more rigorous reliability requirements, eventually we’re going to see modern software hardware capabilities start to address things like time critical safety, critical and related applications, other examples, in the longer term, 5G, and artificial intelligence, blockchain, they’re going to have a greater effect from a scaling perspective in terms of what’s possible with Edge Computing and the automation and control environment.

There’s definitely going to be a transition period here where we can take advantage of what’s already available. And then you’ve got these longer-range technologies that are going to be a little bit more disruptive to what we’re seeing today and their current architectures.

David: Thanks for explaining that, John. I mentioned in my introduction to this podcast, how Edge Computing is well known for its data aggregation and analytic capabilities. Can you give some examples, Aric, of what’s possible through some of the user end user projects you’ve worked with?

Aric: Definitely, that data collection and analytics are the heart of these digital transformation efforts that are happening with almost every customer out there. It seems like right now, these customers want to develop and understand how their production environments are performing versus a benchmark or a golden batch, and do they need data from all the relevant parts of the environment. And that can include certainly the edge but their maintenance elements that can be involved, consumables that are being utilized throughout the process, energy usage, full supply chain and timing. We’ve seen operator variables that can play into these decisions and equations and each, all are augmented by [this data] more and more and being incorporated to these models they have to create, so they can compare things and identify the golden batch and compare current operating environments to that golden batch.

David: What about from an operations perspective, does Edge Computing change anything from an operational or engineering standpoint? Does predictive maintenance – being able to get mobile alerts, mobile monitoring, in some environments, do people want to do be able to go in and troubleshoot and potentially even change some of the things that are happening in their environment?

Aric: We don’t see that quite as often. But I think people are considering how they can do that in light of restrictions on how many people can be in a plant now and things like that, that come along with COVID. Getting health alerts and knowing when something’s going to fail before it does, having redundancy so that when something does fail, you don’t have any issues with your production. And it can auto switch over and have hot swap happen with a failed unit, that and identifying bottlenecks throughout the process are all big advantages operationally.

David: Okay, thanks for explaining that, Aric. And you I just want to note here, you mentioned predictive maintenance, and that’s still a big leap for a lot of manufacturers, considering that many have only recently begun moving from reactive to proactive maintenance processes. So can you explain more about how Edge Computing can help manufacturers move ahead into predictive maintenance?

Aric: Sure, in the Stratus platform, for example, we have analytics that are happening that we can see if something starting to operate outside the parameters that we would like it to. And then we have a redundant setup as well. So you’ve got this predictive maintenance built into the hardware. But there’s also, an example of machine learning, there’s software out there that can be loaded onto an edge device, and can begin to identify when things aren’t functioning the way that they’ve historically worked.

Doing that there on machine really eliminates some latency issues of having to send data that’s been either filtered or just raw data to a cloud or a server somewhere, have it crunched, analyzed, whatever analytics, run on it in software, and then pushed back down to the plant that takes a lot of time and something can go wrong. And if you’ve got a latency issue there, you wouldn’t know and you can still have a downtime event, despite having something loaded that’s not there on machine. That being on the asset is key to this pattern, pattern recognition and other things that need to happen for predictive maintenance to be truly an advantage.

David: Terms like redundancy with auto switchover capabilities, hot swappable failed units, bottleneck identification. They’ve all been mentioned here in our discussion so far today. John, can you explain these a bit more?

John: Sure, the broader term is high availability, and Stratus is fault tolerant, high availability solutions, both our software and hardware based capabilities are designed in such a way to not only provide high availability or fault tolerance, but the important part here is no loss of data. And when in a redundant configuration, when a primary device fails, it switches over to a secondary device and that secondary device becomes the primary with our 24 by seven (7) days a week remote monitoring, the replacement process is essentially autonomous with the replacement device automatically delivered within a couple of days now.

When the bad primary is replaced, it’s a simple hot swap. There’s no downtime there and the new replacement devices installed. The switchover happens. It’s not only self recognizable, but also auto synchronized, and you’re back to the same state, fully redundant, high availability situation that you were in previously. So that’s really the common capabilities that Stratus is actually well known for in terms of both fault tolerance and and high availability.

David: Understood. Aric, back to you. Much of what we’ve talked about here today is from an end user perspective, and considering your focus on OEMs and VARs, what about OEMs? What about these groups? How did the deployment and implementation points we’ve discussed here today, change for OEMs applying Edge Computing to the equipment they sell?

Aric: Sure, so many OEMs are developing edge and Industry 4.0 capabilities and putting it on their equipment today. It’s in order to see a benefit for themselves, give their end users what they’re asking for and those capabilities and win more projects and even, charge a higher premium for their machines. They’re seeing design and deployment gains in time, as we talked about earlier, and they have the ability to provide aftermarket services to their end customers.

On that aftermarket service front, we see many end users asking and looking to outsource these services, again, in light of COVID. But it’s also to be able to focus on what they’re good at, versus some of the maintenance and predictive maintenance that’s happening out there. As they look to do that outsourcing, these OEMs are uniquely positioned to provide that because they’re the experts on their equipment. And they probably commissioned it in the users plant environment as well. So they need to be looking at offerings out there that have either services built in, as [we do at] Stratus. When we sell ztC Edge, it gives them the ability to take that health monitoring as an example, and say, okay, we know that that equipment has something that’s going on that could cause an issue. In the very near future, we can now tell our end user and charge them a fee, so that we can give them that information and keep their UI at a higher level, keep their plant up and running and keep their production higher after we’ve installed and sold the machine. So they continue to provide a value there, those end users are looking for that and as the OEMs build all of this into their machine and consider all these strategies, they position themselves, in a way if they’ve got this capability to win significantly more projects against their competitors.

David: So staying on this aftermarket service topic here, one thing we’ve seen, especially in the past year, in light of COVID, there’s definitely been a trend toward adoption of remote access technologies for aftermarket service applications. But in some industry verticals, you know, consumer packaged goods, in particular, there’s still a lot of resistance to having outside organizations be able to remotely access production equipment, even if it’s by the OEM, who built it. So given that, John, are there aspects of how Edge Computing enables remote access for aftermarket service that helped make these remote connections more palatable to cautious manufacturers?

John: I’m sure there is. You know, first of all, it’s important that these companies have a cybersecurity model and a strategy in place where they’re not only securing their data, and their app algorithms, but establishing trusted communications and computing capabilities in their environment. Additionally to that rigorous authentication, access control and authorization, essentially, what’s called triple A methods either in place by their IT or security office, or through partners’ solutions.

There’s a lot of companies actually bringing new approaches here to the industrial environment. Specifically, with Stratus, we employ a rigorous security development lifecycle process, and how we develop our products. And that includes platform security, host based firewalls, allowing the users to blacklist or whitelist IP addresses, domain names, protocols, reports, we restrict USB ports to essentially prevent unauthorized access, role based access control with enhanced password management and Active Directory integration if that is something you’re employing in your environment. Of course, all data sent between Stratus devices have to be securely and encrypted channels between the communications from our devices and what other sources are accessing our devices. And of course, secure trusted boot ensures that at startup, we have signed and verified bootloader BIOS device drivers and OS files. So what we do to secure the device foundationally allows or should provide some comfort to the end users in terms of our device security to essentially make it more palatable to the manufacturers.

David: So to sum up our discussion here, John, you know, what would you say are the key values that Edge Computing brings to the manufacturing and processing industries as it relates to industries’ digital transformation today? And are there any trends, you know, as a supplier of edge technology, are there any trends you see developing in the near term that might boost those values?

John: First and foremost, we’ve talked about lowering costs, obviously improve margins and developing products and in terms of consolidation, we’ve talked about that more lean and flexible operations, right? The obvious innovation opportunities that come forward with modern technologies, you’re bringing that into their operations into their workforce. And that means skill advancements that enable manufacturers to develop their own skills in house or if they don’t deem that particular competency valuable, they can outsource to integrate us or solution builders, where they see fit. So at the end, you know, really this is about customer satisfaction, improving customer satisfaction, and enabling competitive advantage.

When you start to look over the horizon here, let’s say beyond COVID, I definitely see a lot more cloud-to-edge and edge-to-cloud usages coming forward. And that’s from the cloud service providers, telcos, I think they’re going to start to play a greater role, and the transformation at the edge. From an IT technology perspective, we’ve talked about virtualization, but software defined technologies is the broader umbrella service management, and that’s including fault tolerance, systems management, edge analytics, I believe AI will come forward more significantly over the coming next two years, cybersecurity, as we talked about, and of course, connectivity technologies, that sort of bridge the IT and OT environment protocols such as OPC, all of these are going to play a leading role in this convergence space. So those are the ones I see really playing out more significantly, in terms of directions and technologies over the coming year.

David: Interesting. All right. Well, as much as we’ve seen over the past few years, it sounds like we have a whole lot more to look forward to so thank you for joining me for this podcast, Aric and John, and thanks, of course, to all of our listeners, and please keep watching this space. For more installments of Automation World get your questions answered. And remember to visit our website at www.automationworld.com to stay on top of the latest industrial automation technology insights, trends, and news.

Related Posts