Privacy Policy

The IoT Show S1Ep1

The IoT Show S1Ep1

ALLAN BEHRENS [00:01:04] Hello and welcome to the IoT show. The IoT show is a discussion amongst knowledgeable people in the industry talking about topics of the industrial Internet of Things and the Internet of Things. We’re amongst friends here and the objective here is to talk about topic situations, opportunities and gotchas within the industry and particular areas of implementation of IoT and industrial IoT. So I’d like to welcome our three guests. Thank you very much for coming in. There’s Bill Boswell from Siemens. Josef Waltl from from AWS, Amazon and online we have Diego Tamburini from Microsoft. So let me just let me just hand over to each one of them. They can give you a bit of background on themselves and their organizations, Bill.

BILL BOSWELL [00:01:59] Okay. Great Allan, well thanks for having us here in the discussion today. I’m Head of Marketing for MindSphere on our Cloud Application Solutions for Siemens. MindSphere is a cloud based open IoT operating system. So really excited to be here today to talk about IoT and IIoT.

ALLAN BEHRENS [00:02:16] Great. Thanks. Josef.

JOSEF WALTL [00:02:18] Hi. I’m the Global Partner Ecosystem Lead for industrial software at AWS at Amazon Web services. Industrial software covers all types of partner applications for three value chain steps. Product design, production design and production. And we manage a large ecosystem of partners that help us bring the value of our platform to the end customer.

ALLAN BEHRENS [00:02:45] Great. And Diego.

DIEGO TAMBURINI [00:02:48] Hi, my name is Diego Tamburini, I am Principal Industry Lead for manufacturing in our offshore engineering division, our cloud team, and our team works with the ASHA product development team providing the industry perspective, there are unique opportunities and challenges of the industry so that they can develop the best platform for for manufacturing and of course a big portion of that is IoT. So in order to do this, we we engage with industry and industry organisations, industry influencers so we can learn as much as we can about the industry and apply it to our platform.

ALLAN BEHRENS [00:03:27] Great. Thank you very much everybody. So this show, as I mentioned, is is amongst friends. We are very keen to try and help you, the listener, understand how to benefit from the industrial IoT and the IoT, how to apply it. And some of the gotchas as well. So I think as an opening I’d like to just ask each of the companies, you know, what their view of what is the industrial Internet of Things. It just very briefly. So, Bill, give us your view.

BILL BOSWELL [00:03:56] Sure. Well, I think that the industrial Internet of Things is is really the Internet of Things almost on steroids. Because when you think about what has to happen in manufacturing, it’s really about the context. So it’s not enough just to have single connected sensors or or kind of simple networks what you’re really dealing with is automation systems that are very complex, often brownfield equipment that’s been in facilities for years and thousands of sensors, particularly on any machine or any pump or a locomotive or a wind turbine, anything that you think of. And so the way I like to think about it is context becomes really important for it. It’s not enough to just be able to monitor a signal and say that it moves. You have to really know a lot of domain knowledge about what it is that you’re looking at to be able to make smart decisions with that data.

ALLAN BEHRENS [00:04:45] Great. Very interesting. Diego, how about you? What do you think?

DIEGO TAMBURINI [00:04:49] Well, the definition I like to use is very basic, is a network of ioT is a network of devices that are connected to the Internet and are sending in sensor data and commands to to each other or to and from software systems. And of course, the industrial IoT is where these devices are, basically not consumer devices, but that is industrial devices. And I think that the IoT is really, really about data. I mean, you have these millions of devices that enter the Internet ecosystem generating huge amounts of data that that need to be interpreted and from which you need to get insight.

ALLAN BEHRENS [00:05:41] Great. Thank you, Josef. How about yourself?

JOSEF WALTL [00:05:45] I think building on what both of you said. I think when you have this masses of data from all these sensors, from brownfield applications, from use sensors that are deployed in the field, you really need powerful centerpoint to do all this applications. Just to give you a data point, an oil platform generates approximately hundred terabytes per week. Approximately 10 percent of this is streamed onshore for additional analytics. This is by far more than a lot of consumer IoT applications can handle. And I think therefore it’s very important that we get the connectivity there, but that we also have to write analytics in the back end and that we can do analytics wherever in the world, because all those industrial applications are not only located in one area like a continent or whatever. A lot of these companies that are the customers really operate around the world and therefore it’s critical to have services in place that also can analyze that data globally.

ALLAN BEHRENS [00:06:52] Right. So so let me go onto the next questions and say why is know what is the value proposition for industrial IoT? Why do you come? Why should companies think about the industrial IoT and why is it important to them?

JOSEF WALTL [00:07:07] So industrial companies. First of all work with their with their own large capex. And they tried to optimize the use of the capex and optimize a variety of KPIs the best known is equipment uptime, overall efficient equipment efficiency, but also energy optimization with all the renewables and fluctuating prices comes into place. This is the one part and therefore all sensor data. Basically all data on the status of things out there and equipment out there and additional analytics helps to improve those KPIs of productions. The other thing is business models change. Companies that have sold big machinery like paper mills or whatever will sell paper by the hour or valve operating hours or whatever, and you can only sell something as a service in a buy use model. If you know the status of a system in operation to make it economically viable and this is the next big thing that we see with working with our partners and customers, that industrial IoT really helps those industrial end customers to change the business model to a more service like model.

ALLAN BEHRENS [00:08:31] Diego, I mean, have you got.

DIEGO TAMBURINI [00:08:31] Yeah. And piling on what Josef just said is basically IoT investments help the top line by adding more revenue opportunities and their bottom line by by a reducing waste or reducing expenses. So so a new sources of revenue, for example, attach services like predictive maintenance and optimization services. And that means, for example, predictive maintenance in high capital equipment the savings in predictive maintenance alone justifies the IoT investment very quickly. And then there is also the ability to deliver new customer experiences, connected devices support the new, new and interesting scenarios. And like Josefmentioned, the product as a service model are enabled by the IoT. And lastly, the inside that that that the data is sending, particularly around how your product is being used, what features are being used the most or not used. That provides valuable insight to to prioritize the features in design or, for example, to identify upsell opportunities in you in your sales department.

ALLAN BEHRENS [00:09:53] Great. Very interesting. Thank you, Bill.

BILL BOSWELL [00:09:55] Yeah, I think that it’s really a lot of interesting possibilities. And when you look at the use cases, you know, it goes beyond just the obvious monetary use case as well, you can imagine what happens if you have to take a train out of service. It’s already running on the line, right. Because a door didn’t close properly. So you want to be able to do things like condition based maintenance and understand what failures are going to happen and be able to schedule returns or repairs when the assets not in use. You know, the use cases that we’re seeing across industries are are are really just amazing. Even the term smart maintenance means different things in different industries. And so that domain knowledge that we talked about, whether you’re looking at an automation system with PL Caesar, you’re looking at mobility systems or smart grids who have energy management, it means different things to different audiences. And but we see IoT being adopted across all those all those areas. And it really comes down to I think that the ability for people to be able to see a place to get started. Right. And get started fast on that.

ALLAN BEHRENS [00:11:01] Yeah, fascinating. So Diego I mean, let’s talk about how how one goes about justifying the investments into IoT. Do you see any common themes or is it very predictably verticalized? I mean, give us some sort of feedback on that if you can.

DIEGO TAMBURINI [00:11:17] So, yes. One thing is done on that estimate the complexity of an IoT project. But before starting an IoT project you, my recommendation will be to focus on on one and only one business problem. So have it really clear what is that you’re trying to solve or improve and then identify that that the key KPIs that that that will let you know if you were successful solving that problem. So so then when you when when you do that, the next step is to to identify what is the data that you need to solve that problem. And start the IoT project by using simulated data first, real device come later. Do a proof of value, see that you are successful and start with small and and and incrementally grow the complexity and the number of devices and the number of standards, et cetera. Start a small start easy. Identify one problem that you’re trying to solve.

ALLAN BEHRENS [00:12:30] Sage advice. Josef, if you got anything to add to that or.

JOSEF WALTL [00:12:34] No. Our experience is very much along the lines than what Diego has just said. It is important to start small. What we also see that it is important also to get a data quick because it’s not in a way that you build up an IoT system and then you have all the predictions, predictions need data therefore start to collect data even while building up the whole system and analyzing algorithms. And the other thing is what we saw is start with small teams. In Amazon there is this saying of two pizza teams, approximately eight people do not start build big organizations for defined problems and to sprints like six to eight weeks. And you can also do it in power. We’ve very had very good success with success with customers and partners that had three, four teams working in sometimes also a bit competitive matter and then stop, see what you’ve learned and then reconcile and go for the next sprint.

ALLAN BEHRENS [00:13:35] Great.

DIEGO TAMBURINI [00:13:36] If I may I add something, too. Josefmentioned something really important that if you are planning to use a IoT data to do any sort of predictive analytics or predictive maintenance, these things don’t work from day one magically is people, don’t sometimes it means the fact that that this particular machine learning models need to be trained with a lot of data and in order to do that. That’s probably the hardest part of an IoT project that is using predictive analytics is if, say, for example, if you’re trying to predict failure, you need a lot of data that that illustrates when when failure happens and you need to do to identify what are the same sort of parameters that are important for failure and even what failure means to you to to train these machine learning models. They don’t they don’t work from day out, day one when you turn on the system.

ALLAN BEHRENS [00:14:39] Okay, Bill.

BILL BOSWELL [00:14:41] So I think one of the exciting things, though, is that, you know, as we look at these cloud based IoT systems, we’re able to help companies get started fast for that. So, you know, you don’t have to be a large company with lots of data scientists and programmers to work initially on the machine learning. We’re seeing small and medium sized companies being able to hook up assets today that are already support OPCUA as an example or other protocols and be able to really just from the condition and monitoring and alarming and learning with simple thresholds are able to realize the big benefit to that and then move into what do I do now with more of the machine learning and the AI types of application? Of course, that that’s where I agree 100 percent with what the previous speakers have said about start with the end in mind and make sure that you know what you’re analyzing and what you’re trying to accomplish with the data. But the neat thing about the cloud is, is that small, medium sized companies are able to get started with a lot of out of the box applications that help them move much forward more quickly than they could have if they were trying to develop it themselves.

ALLAN BEHRENS [00:15:51] Just to sort of add two penneth, actually, there’s a number of companies have talked to me about pre populating the datasets that are used for these types of machine learning environments. So that also helps, you know, bypass some of the early data gathering stages.

JOSEF WALTL [00:16:09] Yes. I think I’d like to add something on what Bill said, because I think it’s really important for a lot of companies that need to make decisions. The one thing is, uh, companies need to understand where their core strengths are. Core strengths can be build a better robot, build a better paper machine, build a better oil rig. And then there is all this analytics on top of it where they can get better in operations or change their business models. But having said that, the more companies build up from from scratch, the more they need to get softer companies. That means they will have internal departments that will have data science scientists. And this needs is a change in the company itself. This is where we work a lot with customers to help them. But on the same time, there is a lot of partners from from from us and from from other companies that specify in specific problems in the industry, like predictive maintenance on big motors or whatever. And this is standardization like in any other like in IT history basicall that drives a lot of value for customers. I think that needs to be answered for each company on a strategic level, what do I want to be, do I want to be an analytics company. Do I want to compete there? Do I want to use software that makes it easy for me to get to sales fast? And then very often we see hybrid scenarios where part of the get go and the start is done with a partner also with existing connectivity based on hardware some of these partners provide. And then when it comes to specific things, those companies say this is something that I can think I can do better than any other company in the world. And then they build up their own competency.

ALLAN BEHRENS [00:17:58] I mean, Diego have you got any thoughts about important learnings that you’ve seen or your your customers have seen that you want to share with the audience?

DIEGO TAMBURINI [00:18:12] Yes, sure Allan. One is the, Josefalluded to that a little bit is that figured out if you if you are trying to to make money or save money? So it’s that. What is your business model? Are you trying to optimize your factory operations? Are you trying to reduce the energy consumption in your factory? Or are you trying to deliver new services and new experiences and a one learning and something that I heard a lot a couple of weeks ago at IoT World in USA San Diego is that pricing is very hard to predict a cost how much you write IoT solution is going to cost. So particularly if you are architecting your own solution, any how you architect the solution may make a huge difference in the operation. cost of your cloud IoT solution. So so a plan for that. Don’t underestimate the pricing calculator. I’m sorry, the calculation. We we offer some calculators that that help you estimate if you are using these three services with so many devices. This is roughly what is going to cost you per month. But if you are architecting your own solution. Keep in mind that that some solutions like where you are going to store your data. How are you going to do your stream analytics? Are you going to use distributed databases or relational databases? Does make a huge difference potentially in the in the cost of your solution.

ALLAN BEHRENS [00:20:03] Right. Right, Bill.

BILL BOSWELL [00:20:05] No, I think that’s a great point, Diego. You know, I have to kind of link back to thought that you said earlier, don’t underestimate the complexity of it. That’s one of the things that I think we get excited about working with partners like AWS and with Microsoft in that, you know, by building the platform on top of the public cloud providers, we can help steer companies into what do they want to be. What is their core competency? Right. So by providing access, both that kind of the managed service level and then directly to the native cloud environment as well. You know, our our goal is to kind of help keep the guardrails up, to help companies make decisions that keep them in the boundaries of that. And that way there’s access to both the native cloud technology into the managed services as well. So I think that that’s a great point that Diego made, is there is a lot of complexity, but with I think the ability to get started fast and with a defined problem and try it fast, you can learn a lot about what it’s going to be like to implement IoT in your company and in a pretty straightforward fashion.

ALLAN BEHRENS [00:21:07] Right. Josef, any other sort of thoughts on these new answers?

JOSEF WALTL [00:21:12] I’d like to to even take further what Diego has said on the on the cost and on the prediction of the cost. So industry has big assets. Big decisions tend to lead to more conservative decisions. And there first of all, companies say I’d like to have my own environment. That means that there is a certain types of fixed cost even in a in a in a cloud environment there. And what we see is that the more cost is getting affected and it will get even with industrial IoT applications, more and more effector people move to more a service based architecture where you consume a service like you do in other IT fields, like CRM already, and everyone is fine with it, because this allows that the underlying infrastructure is shared among several customers and the individual data point is getting cheaper for the end customer. And this does matter, especially when we go into the SMB field. This is where we work closely with partners like Siemens to really optimize the architecture so that the best use of the existing assets that technology horizontal technology providers like our company can give to drive the cost for the individual end customer to a reasonable level that it makes sense for its own specific business.

ALLAN BEHRENS [00:22:41] Yeah. So so more applicable to the SMEs. Yeah.

ALLAN BEHRENS [00:24:11] So what about gotchas? I mean, we all know that with any emerging technology, we’re learning very fast to failing fast as well, which is a good learning experience of course.

BILL BOSWELL [00:24:22] Why you pointing at me.

ALLAN BEHRENS [00:24:25] No it’s just what you said, Bill. So we we need to learn fast. We fail often and learn from our experiences. So what are some of the gotchas that we’ve that you’ve heard about or or things that people have been surprised that during the IoT journey.

JOSEF WALTL [00:24:42] I can give you an example where we work with a partner customer that rightfully for its audience of customers in its home country, build up an own managed service business with its own cloud. And then the first customer came out of Singapore and then we get into discussion and then helped really to to serve this customer with the technology of the company. But based on our global infrastructure, that’s a thing. When when you work, even with companies in your home market, the probability that the application is somewhere where not you’ve never experienced is quite high. So think first, really on a global scale business model because of especially manufacturing facility stand all over the world. And this is where the data will come from. Therefore, you need to have a plan for that.

ALLAN BEHRENS [00:25:37] Right, Bill.

BILL BOSWELL [00:25:38] Well, I think that’s a great point. I think you know, one of the things that’s really interesting is that before people really didn’t have a chance to have that kind of global view. And there is kind of the the the law of physics as well that comes into play here is there’s always gonna be a hybrid architecture in the industrial Internet of Things where there’s going to be work that you’re doing on the edge. Right. Analytics that you’re doing on the edge because it has to be completely real time. And and because of maybe, you know, latency up into the cloud, there are things we have to take into account. But for the first time, we’re seeing customers being able to bring in information about what’s happening on that line in that factory and what’s happening at the same production lines at five other factories and be able to look at the overall picture of what’s happening. And they really do need to be able to scale that globally and be able to operate in a in a global environment like that. And so the scalability is not just the amount of data, but it’s being able to support the different laws and the different different restrictions that you have by operating in different countries. So that’s one of the things I think that back to your question of lessons learned that perhaps people need to think about. It’s it’s you really do want to start with a small project that you understand what success means. But you really do want to think about as you’re thinking about platforms, about what are the overall requirements for your company and where do you need to scale to both in terms of the amount of data and countries that you’re going to operate in. Because while while IT companies and great providers like their partners are going to have experience in the software side of that, if it comes down to supporting the automation system to the factory or plant or city, it’s you really are going to need some some feet on the ground and some assistance to be able to help with that in the countries around thw world.

ALLAN BEHRENS [00:27:28] And the Segway for the for those watching the next episode’s on platforms. Diego. How about yourself?

DIEGO TAMBURINI [00:27:36] Yeah, I think consider the possibility that you may be successful with your IoT project and that that that you’ll find yourself ingesting terabytes of data a week like Josef mentioned earlier. And so consider the possibility that you’ll have to scale very quickly and then you’re going to start getting customers or deployments all over the globe. So you have to concede like Bill was mentioning data storage regulations. There are some restrictions around that and governmental regulations, certifications on the cloud platform that are required. And then the latency Bill also alluded to this is the round trip from the device to the cloud may not be acceptable in your scenario. So you would have to consider a processing closer to the device and the other things that is almost comical is that I hear over and over again is the friction between IT and OT, the information technology and operational technology. So IT the folks that connect software assistance basically and OT are the ones who connect machines and they’re supposed to be working together. So IT often perceives OT as reckless, naive. When it comes to security and OT perceive IT as obstacles, people who delay that implementations with red tape. So so there is an organizational issue that that you have to address early and of course IT always feels a little bit threatened when it comes to cloud solutions. So so I think that that needs to be addressed early. If you have distinctive IT and OT departments.

ALLAN BEHRENS [00:29:29] Great. Thank you.

BILL BOSWELL [00:29:31] Can I add one comment on that, but I think that’s a great point, Diego. And you know, the other thing I think that that you want to do is plan for success, as you said. Right. Don’t be surprised if you’re successful, because the other thing you’ll find is once you begin to knock down those barriers between organizations and it’s not just IT and OT, it’s manufacturing and engineering and and other parts of your organization that could actually make value of the data that’s collected about the digital twin of the performance in the field, being able to tie that back into the digital product twin and the digital production twin as well. There’ll be many people who will be able to make use of the data that they never had access to before, and they’ll bring other data sources to be able to combine that into and be able to do big data analytics on it as well. So as soon as that data becomes available, all of a sudden there’s a whole new set of challenges of people that want to use it. And so you need to plan for success in that arena as well.

ALLAN BEHRENS [00:30:23] Great. Right. So I suppose, you know, in sort of bringing this to some form of conclusion, there are obviously going to be a lot of companies that sit on the fence. You know, is this the right thing for me? You know, should I be looking at this? I have maybe haven’t got the resource. I’m an SMB. I haven’t got people in software domain who can help in this. What would you say to those people who are sitting on the fence, Bill?

BILL BOSWELL [00:30:44] I’d say absolutely get off the fence, get started, try something. You know, it’s it’s so easy today to go and, you know, sign up online connect devices that are already supporting protocols that are coming out of it. And just, you know, it’s it’s since we’re doing it in the cloud, it’s not a big infrastructure cost. So you’re just able to get on and try it out, get on and learn. Have the project success criteria in mind. But there’s really the biggest risk that you have is waiting because it’s the opportunity cost, because I guarantee, if you’re not looking at it of how you’re doing it, your competitors trying to figure out how to be more efficient or how to implement those new business models or bring in new sources of revenue. So get a team assigned to it, get a champion internally and get started.

ALLAN BEHRENS [00:31:31] Diego.

DIEGO TAMBURINI [00:31:33] Yeah, I second what Bill just say get off the fence. Otherwise, your competitors will. Before you and back to the comment that we’ve mentioned, over and over again start small, identify a small, agile team of people who are passionate about IoT and do almost like a little scrum team with people with a software electronics industrial, and pick say if you’re trying to optimize your factory operations pick one machine a couple of communication protocols to get your blood flowing and have success with that and then grow and grow and in baby steps.

ALLAN BEHRENS [00:32:23] Great, Josef.

JOSEF WALTL [00:32:25] So a lot of things already said, I think just two things to add from my perspective. Do not believe your technology architecture that you set in the beginning to be there for the next 10 years because all of the innovation that comes from cloud providers will allow to grow and facilitate a better platform over time. If you have a micro service architecture that is loosely coupled. No problem. You can even after a later point in time, bring this additional innovation into your applications. And the other thing is for the part where you want to grow the top line, get new services out. Also, talk to one customer or a couple of customers that you know you trust early on. Make sure that you are really helping them and also try to be a bit flexible in what you will provide. Because technology is really there finding the customer problem with all the data of the devices that you can now have with IoT technology is the real challenge and you want to be there before your competition is there.

ALLAN BEHRENS [00:33:29] Great. Thank you very much. So, Bill, Diego, Josef, thank you very much for your time. I think we learned some really interesting things. Well, I certainly did. And I think we are seeing a rapidly evolving ecosystem and set of technologies. Appreciate your time and I hope that you found it interesting. We’ll see in the next episode which is going to be on platforms.