In episode 101 of the IoT For All Podcast, Evan Cummack, General Manager of IoT & Wireless at Twilio, joins us to talk about device security and innovation in the IoT space. Evan shares Twilio’s approach to security through a device’s lifetime, his advice for companies getting started in the space, and what it’ll take to get IoT innovating at the rate of other leading technologies in 2021.
Evan Cummack is the General Manager of IoT at Twilio leading product, engineering, and go-to-market functions for the business unit. Evan joined Twilio in 2011 as one of the company’s early employees and is credited with building several of Twilio’s first products, consulting with Twilio’s largest customers on their global communications solutions, and launching Twilio’s IoT business.
Interested in connecting with Evan? Reach out to him on Linkedin!
About Twilio: Twilio is the unseen infrastructure powering communication behind the scenes. Chances are you’ve interacted with Twilio, you just don’t know it. If you’ve ever texted your Lyft driver, logged into your Netflix account using a texted security code, or texted your Airbnb host, you’ve used Twilio.
Key Questions and Topics from this Episode:
(00:35) Intro to Evan Cummack
(01:21) Introduction to Twilio and what prompted Twilio’s move to IoT
(12:06) Can you describe Twilio’s approach to IoT?
(15:30) What can be done to improve the rate of innovation in IoT?
(19:48) What do you think about when planning for the security of a device throughout its lifetime?
(22:59) How do you advise companies in choosing between full-stack IoT development platforms and building it themselves?
(26:10) What approach have you taken to security firmware updates? What have been the challenges there?
Transcript:
– [Announcer] You are listening to the IoT For All Media Network.
– [Ryan] Hello everyone and welcome to another episode of the IoT For All podcast on the IoT For All Media Network. I’m your host, Ryan Chacon, one of the co-creators of IoT For All. Now, before we jump into this episode, please don’t forget to subscribe on your favorite podcast platform, or join our newsletter at iotforall.com/newsletter to catch all the newest episodes as soon as they come out. So without further ado, please enjoy this episode of the IoT For All podcast. Welcome Rohit to episode 19 of the IoT For All show. It’s great to have you and thanks for taking the time to connect with us today.
– [Rohit] I’m delighted to be here, thank you for inviting me to the show.
– [Ryan] Absolutely, one of the best ways for us to kinda get this started would be to provide a short introduction about yourself and if you can give some relevant background information, anything you think the audience may find interesting, that’d be a great way we can start.
– [Rohit] Sure, so my name is Rohit Tripathi. I am a senior vice president, head of products and Go-to-Market for SAP Digital Interconnect, one of the end to end units within SAP. As a background on myself, I’m an engineer by training, worked in supply chain space, went to business school, after some stints in strategy consultant landing up in SAP through various roles. And now I’m running the products group here. And I guess that’s got us into some of these discussions with you guys on IoT and other tasks.
– [Ryan] Yeah absolutely, I think, would you give a little bit context in kind of how you got into the space? I know, just doing some basic research you came from Boston Consulting Group and you have a background in that area. How did you kind of get into IoT, get into the role you’re in now? What kinda triggered that change, or that at least evolution into there?
– [Rohit] Yeah, I mean, I think, if you think about it, right? Like from my supply chain days, right? Like the inventory tracking, shipments, managing supply chain activities is a challenge, right? It’s something that we still face today in businesses. The first step of that was putting in enterprise software to address those concerns. But yet the real-time nature of that was limited. And I think in early 2000s if you guys remember then RFIDs were a big thing, that’s when we started doing that to get some real time signals to address these things. And I think that what led me to get drawn to this whole machine to machine conversations, things to machines conversations, and which then now or time it has a evolved into what we referring to as IoT today. So I think it was that background that led that interest.
– [Calum] Could you talk a little bit more about supply chain? So that’s obviously a big area in IoT, but I’d be really curious to hear about your experience and what you’ve seen since you’ve been involved, or were involved in that several years ago and have likely seen a lot of changes there as far as new technologies, enabling different applications. And when you first got into it, what was the cutting edge? Whether it’s RFID or that enterprise software you mentioned and what does it look like today if you’re still involved at all?
– [Rohit] Well, so in the very early days, right? And then even I was actually just barely getting out of college. So I’m not that old guys. So being that in. So in the ’90s, very early ’90s, if you think about what was happening in supply chain and the manufacturing world, that was the first wave of digitization, so to say. And essentially at that point in time, the basic materials planning processes were being digitized. The processes were being standardized and that’s when the big ERP providers like SAP and others started coming into like being relevant and providing clear impact to the organizations. The next level then that grew once you had those processes standardized and material planning things in place, in the 2000s, the big thing was all about supply chain, planning and optimization, And that’s where RFID and other elements started becoming relevant. I think the challenge then that these softwares and even the customers, you of them, faced was, that the hardware, the in-memory capabilities were not enough, or were very expensive to make things more real time. And I think now things have become more real time. Like we all expect things to be, information to be updated instantaneously, right? Like, of course there is some lag, but it’s no longer acceptable for any manufacturing organization to say that we work on plans that are like one week old or two weeks old, or information is as of last month, right? So I think that’s the biggest change and I think that’s well connected within IoT are like leading that forefront now if you think of the manufacturing digital supply chain.
– [Ryan] Yeah, I was gonna say that to me one that I would guess would be the connectivity piece. So if you’re dealing with supply chain, that’s often global. And so if you’re tracking–
– Right.
– [Calum] Thinking over very large geographies, that means you’re gonna need certain kinds of connectivity. So like recently there’s been a lot of an upsurge in IoT applications and things using like low-power wide-area networks like LoRa, but–
– Yes.
– There’s a and so you have to install those in a specific area. Whereas if you’re going–
– Yeah.
– Over borders or across states, you probably need cellular connectivity, or perhaps even satellite if you’re going to remote locations. So what are your thoughts there as far as like the evolution of connectivity and how that’s been impacting supply chain, or other Applications or applications you’re involved with at SAP?
– [Rohit] Well, that’s really, if you just, you mean I could not agree with you more, and I think if you think of the current supply chain today, right? Like you go, let’s go from a factory, right? Like that’s where things are built. Then they go to a warehouse where they are stored and then broken up into pallets and loaded into things, and then they are shipped, right? If you think of each of these three domains, connectivity is needed, but they all come with its own respective limitations or challenges. So when you’re in the factory, actually, because it’s fully under control, WiFi is actually a very reliable use of connectivity there, right? But then very quickly, one needs to switch over when it comes to a warehouse and you don’t have more geographical disparity coming in, but it’s still controllable and that’s where things like LoRa become quite relevant as you are referring to. And you’re spot on, right? Which is warehouse environment is perfect for those kinds of low-power connected options. But then once you put them on trucks, which are then crossing across state lines, or even when you think internationally they are crossing across country lines, or ships, then perhaps cellular and WiFi is the only option. And I think today in supply chain when thinking of connectivity this is something that one needs to, like for those of us in this business who are evaluating options of connectivity, they need to make sure that they address all three of these, right? And not just focus on one or the other, more importantly, I think people should think of connectivity when thinking of IoT. I don’t think many people think of connected with you when they think of IoT.
– [Calum] Right, and not only just thinking of connectivity which is a critical piece of it, but you know what connectivity makes sense. And is that the only one you’re gonna need? So, as you were just saying, if you need, the reason you need different kinds of connectivity for different contexts is because there are trade-offs. And so something like cellular is great for certain applications because you have a lot of the infrastructure already built by these cellular carriers–
– Right.
– They can transmit over large distances and can be very high bandwidth. I mean, particularly right now with 4G and with the coming of 5G even more.
– Oh yeah.
– But often the drawback is power consumption. And so you have devices and cost. So if you have thousands of sensors that you want to deploy somewhere and they’re very low bandwidth, they don’t need to let it send a lot of data. Maybe it’s a temperature reading or soil moisture, then for those maybe cellular it doesn’t make sense. But in the supply chain example, I find that really interesting because you could potentially have things that are moving across these different spaces, where they are in a manufacturing setting. So one connectivity type makes sense, but then they move and now they’re being transported. And now that connectivity type might not make sense. So it seems like a core piece is you’d want things that can adapt to different kinds of connected pieces. Would you agree with that?
– [Rohit] Yeah, I mean, I think the element is that if, it’s interesting that like as we are discussing through this, to some of our listeners it may appear that, oh, this is complex, right? Like we have already started to what was a simple us which was just connect my IoT device, right? Like we already making it sound that the complexity is increasing. So in a way, yes, we are bringing in different formats. But essentially what we’re doing is, what we are saying is that one needs to think of these various dimensions of connectivity for the, actually the whole or successful execution of the IoT investments, right? And I think the element has to be not only looked at from what makes sense in that environment, but then also perhaps from cost perspective, right? Like, because, and cost, you touched upon power, which is one element of cost, but then there is of course always the direct cost of data that flows, right? And then the last part is thinking about security for the device, as well as the data that is flowing from the device, right? Let’s not forget that. And so various protocols will have different levels and layers of security that can come in. And I think that’s also the dimension to think of when looking at these options of connectivity and yeah, I mean, you’re right, man, it has to be these different formats. There is not one answer to say, go do this and then you are connected for your IoT world.
– [Ryan] As I said, can you talk a little bit about then the IoT Connect 365 offering that you guys have, ’cause I think that kind of plays into this conversation pretty well.
– [Rohit] Yeah, I’m happy to. So just before we get into like why IoT Connect 365 and why are we doing this as SAP Digital Interconnect, just a background on the group. So SAP Digital Interconnect is we are in the business of providing last-mile reach and connectivity for various SAP applications. And then also for standalone non-SAP applications as well, like whether it is connecting businesses and customers to their employees, or connecting businesses and things to each other, or to the end customers, right? And I think our mantra is intelligently connecting everyone, everything everywhere. And so what we were doing to everyone with our various channels, and then now with rise of IoT with IoT Connect 365, we are bringing the same element of connectivity to businesses, right? And our objective is that we wanna hide the complexity that we just discussed from our end customers, right? Like we want to say that we be the one conduit to the end customer, but then on the backend, we deal with these various connectivity issues that are there, right? Like much like today, when an enterprise sends you an SMS, the enterprise doesn’t really care which telecommunications provider that the end customer is on, right? They just want the message to be delivered. And so we helped them achieve that, right? Like we take care of the complexity and working on the backend to our APIs with various telco providers and then expose one single API to the customer and to the enterprise, I mean. And so that is something that helps them simplify that communication and we bring the same benefits on the IoT side with IoT Connect 365.
– [Ryan] Right, yeah, I mean, I think one of the things that might bring, kind of provide a little more context to what it is that you guys offering does as we could talk about some in individual situations or Applications that this is actually being used in now. I know you probably can’t mention too many of the customers or clients per se, but at least talking about what kind of Applications are benefiting from the service and kind of what was happening before the service was installed and now using the service the kind of the benefits that they’re seeing.
– [Rohit] Yeah, I mean, there are actually, we’ve got several customers across the globe, so let me talk one about in North America, this is a car manufacturer, right? And what they were deploying is, they were coming up with their new car models and what they wanted to get was some sort of connectivity options in these cars for getting information on maintenance, infotainment and related analytics, right? Like they they wanted these test cars that they were rolling out to communicate that back to their main systems and collect that information. And so, as you can imagine, what they want to find was connect with the option that works across geographies, these are cars, they will be driven, and when they are giving them out on lease to the end customers, you cannot really tell the customer, hey make sure that you always have it connected on WiFi, or only drive in this city or et cetera. So they wanted something that was more global. They didn’t want to deal with like only this operator or that network, et cetera. So they came to us and what we started with them was like, let’s do this in small steps, right? Like, rather than going and saying, let’s install this in the entire fleet, start first with like a small proof of concept. And this started with like 40, 50 cars, deployed the eSIM devices, got the BI integration and sort of went live in the lab, right, so to say. And then they started collecting this information. And then within three to four months, they actually got enough comfort with the service that actually the quote that we got from them was like, it just works, right? And then they went live deploying it across their entire fleet. And now are live with the product, right? So in this case, the IoT Connect service is actually helping them, for them the IoT device is their car, right? It is helping communicate back information that they are taking back, right? It’s a very interesting thing that not many people think of, cars as IoT device. However, if you ask one of my colleagues who drives a Tesla and he actually says that it is nothing but a mobile device with LTE connection. So that’s what he refers to that. So it’s because there’s a battery pack, it has a mobile device and it has an LTE connection. So it’s an IoT device.
– [Ryan] Absolutely.
– [Calum] Yeah, so there was something you mentioned I wanna touch on which was how you went from the proof of concept stage before moving on to a full deployment, and I think that’s really important to emphasize, something we’ve seen from our experience is that by going through that proof of concept, not only can you help well prove the concept and prove value to the customer, so that they can feel comfortable about moving forward, but there’s also a lot to learn in going through that process. So you can make refinements and adjustments as needed so that when you go to full deployment you have the highest chance of success as possible. But something I wanna bring up and my question here is I was reading through the connecting the internet of things white paper on SAP site. And one section I really liked was the eight must ask IoT connectivity questions.
– Right.
– And one of them was, are you prepared to scale? And just to read a quick excerpt it said, “Investing in short-term solutions that do not scale “has prevented enterprises “from meeting their future business needs. “For these organizations, lack of scalability “has caused significant disappointment “and caused them to fall behind their competition.” So it seems to me like there’s this balance where you want to be doing these proof of concepts to be able to, again, prove the concept before moving forward. But you also wanna do it in such a way that you’re then able to go to scale, where it’s not just a proof of concept that never makes it beyond concept stage. So how do you think about that? I mean, really any thoughts you have there, but just interesting hearing you bring up the proof of concepts, but then also going on to full deployment.
– [Rohit] Right, and I think, there are a few elements that we actually encourage customers to test out truly in the proof of concept stage, right? Like let’s look at the first lens of the technology, right? Which is we encourage them to say, go for a solution, go for an approach that allows you to somewhat future proof your investments, right? Like, which is that can then create the solution that you’re deploying today, which let’s say works on the cellular network, will that in the future work on the more local WiFi, or LoRa, or Narrowband IoT networks as well? Can your investments deal with that, right? Because that’s the first question that we encourage the customers to think about it, right? Like how are you thinking about five years down the road, right? At least a resemblance of what you think this could be, or how could it be used, or deployed. So I think we encouraged them to think of that dimension. That’s the technology part of it. Then there is the element of cost, right? Which is how are you structuring the information that you’re gathering, right? Like versus the cost that is required to gather that, right? It’s to think of that scalability, right? Like what works beautifully in a proof of concept because you are collecting very valuable data and creating cool analytics, but then when you roll it out across the entire scale, the cost just escalate and now suddenly the question becomes like is it worth it, right? Because it just looks so good in the proof of concept because you’re testing it out in maybe 50 units, but now scale out to a million units, is it worth collecting all of that information? And if so, then are you using it on the right format, right? So I think that is the other element of scale that we encourage the customers to test their hypotheses on. And then finally, there is the element of scale, which is geography, right? Like, what are their plans? Do they see themselves being say North America focused approach only, do they envision some things in the future that is more international? If so, where? Because those bring in other constraints not just connected, but then also regulatory constraints. So I think those are dimensions that we encourage people to look off or think of when they think of scaling.
– [Calum] Yeah, I’m glad you’re doing that, ’cause from our experience we’ve seen what happens when those kinds of questions aren’t answered, where something that’s really important to understand, and if listeners out there are implementing or even building IoT solutions is that it’s impossible to see into the future. And so to some degree, you’ll never be able to know, but setting yourself up in such a way that you don’t go down a path where it turns out that either cost or technology is prohibitive from moving beyond those early stages into what you might have invested a lot of effort. So as early as possible trying to do that work of all right–
– Right.
– What are we actually trying to accomplish here? And where do we see this going? Is this gonna be relegated to just this tiny geographic area? Okay, maybe we don’t need cellular, maybe we can use LoRa to cover it.
– Right.
– But if you think we move beyond that to something that is more national, or even global, are you prepared to install all of the infrastructure of all these LoRa gateways to be able to handle that maybe at the beginning you should be building what something like NB-IoT, which is cellular IoT or maybe even something else.
– [Rohit] Right, and I think the other interesting part in this is that like, what I often find is that when people are thinking of IoT and deploying IoT strategies, I think in general and rightfully so most attention gets focused on, hey what kind of sensors we will need? And then the next question goes on is like, what’s the analytics platform that we are using? How are we crunching that data? What data are we collecting? What insights can we get? Can we do some predictive maintenance? Can we do AI and ML on it? Everybody gets focused on those. And conveniently people forget how does the data go from the sensor to this platform? Right?
– Correct.
– [Rohit] And that’s what do they got? I feel this, I don’t know and then something will happen, maybe it’s WiFi, right? And then I think that’s where you have to actually encourage the customers to say, or people who are starting out these say, no, that’s equally important, think of that as well, because that is an equally important contributor to your IoT strategy, how are you going to get those sensors to connect to that central platform, but you’re doing all of the school stuff?
– [Ryan] I’m curious to hear a little bit about how active you guys are on your side in the proof of concept phase, kind of helping clients work through all these problems. And if you’ve encountered situations where the customer has not maybe thought through enough on the connectivity side, specifically about scale. So for instance, they choose one technology to build a proof of concept and then the right scale, or if it works but they wanna scale and then they go wanna go larger and they don’t have that infrastructure in place. In that situation, what kind of advice do you have for companies who kinda make that mistake through their planning phases?
– [Rohit] Well, one is that it does require a lot of convincing, right? To people because it’s amazing–
– Right.
– That how much one gets to tend like, fall in love with their own ideas or concepts that they failed to see, or they need to add something else, right? To improvise it. So, very simple example, right? Like let’s talk of connected vending machines. You will say, oh, they are, yes We have the solution. They will all located in a building. Building has perfect WiFi. They’re connected over WiFi. And that’s how they will work. I have my connectivity figured out, great. How do you reset that machine remotely?
– Right?
– [Rohit] Suppose the WiFi drops or something happens, or you have to reset that machine. How will you reset that machine? You cannot have, the plan cannot be, because if you have to scale and you have to imagine you have like hundreds and thousands of these machines deployed in buildings across the city, across the state, you cannot have people go there and then manually reset the WiFi passwords on them or things. You need to build–
– Right.
– [Rohit] In a SIM in there, right? Like for these off one-time activities, right? You may not use it all the time, but you need to have that option, right? And I think it’s pointing out some of these kinds of scenarios. Then you can see the light bulbs clicking and saying, aha, right? That’s why we need to have this other forms scalability also built in, right? Or people will often go in and say that, oh yes we have this organization, we have SIM based connectivity, and across our company, we have a contract with this telco provider and we are going to go with them. So we have deployed their SIMs, we’re done, thank you. Makes sense today, three years down the line that telco gets acquired by somebody else, or your company shifts allegiances, or through procurement processes to another telco provider. Now, what do you do with all of those SIMs that are installed in all those air conditioners that are hiding up in the various attics of your organizations like who goes and replaces those SIM. So you need to have, think about a connectivity provider that actually, a neutral SIM, right? Like that can then be moved conveniently. And so I think, it does require some of those kinds of situations where people do then come back and say that there are options that exist. And certain times it’s also that people are discovering, or they’re actually if you think of the digital world they’re thinking with different business models. So they’re also trying to see what works and what doesn’t work.
– [Ryan] So if someone listening to this is thinking, okay I understand the challenge of I don’t want to commit to a particular provider, or run the risk of any of those situations happening that you mentioned, whether it’s being acquired, or whatever it may be. How is that avoided? You mentioned earlier briefly eSIM. Do eSIM solve this? If so, what are they? Like what are some of the approaches that someone could take to mitigate this risks?
– [Rohit] So one is that I think, think of there are various dimensions that I would say that people have to think about it, right? Like one is that you have to make sure that you hopefully are investing in a solution that has a global reach and scalability. You’re thinking of solutions that support, that don’t tie you down to a particular connectivity provider, right? Like that are more global. eSIMs help because eSIM’s nothing, but, a digital SIM as opposed to like the actual physical SIM cards that we in insert in things. And then now, like for example, the latest iPhone now comes with it, or has the capability to install it. And then there are certainly more easy to program and adjust, but then also being SIM neutral is something that can help on the cellular side. But then also people have to look at things that can help go across, LoRa, Sigfox, Narrowband IoT options, right? And then finally, don’t ignore the element of security, right? Like how do we make sure that we have carrier grade connectivity, network based protocols and firewalls to make sure that we are providing that protection. So I would say these are the four dimensions, right? And just on that point, I think that’s something that we kept in mind as we were doing our SAP IoT Connect 365 offering because we wanted to make sure that we are addressing these dimensions in this context.
– [Calum] Yeah, so as a follow-up to that, it seems to me that there’s this balance that needs to be struck in terms of having customers excited, which is good. And doing many of the things that can be done with IoT to digitally transform businesses, but also serving as a partner and an educator, and maybe sometimes pulling on the reins a little bit and saying, it might not be good to rush far long into this. There’s some things to consider. So how do you balance that? ‘Cause that’s been a challenge for me personally in certain situations of both fostering excitement, ’cause there is a lot to be excited about, but also being realistic and yet not coming off to the customer–
– Right.
– [Calum] As like I don’t want to do this, or maybe they then go work with someone else who isn’t as honest on some of the challenges. Is that something you have experienced?
– [Rohit] Yeah, I mean, I think, and for us that’s why we encourage one in saying that start small, right? Start small and grow with us, right? As opposed to like go in with like commit now for like the entire thing and go in deep with like millions of dollars of connect with these secured, et cetera. We just look forward and say, look, if you’re already coming at that level, then you already solved your challenges and you know what it is, then the only discussion point is about what is the price points can you get and geography and coverage, right? But that’s where not most people are, they are all in the starting point. So we encourage people to start small and grow. And I think the best thing that we have found in that option is where we work with them, give them these options, and the only way to to help ease this is to actually look, a whole bunch of these processes connect with the options, et cetera, going digital, and to go along with it, what we do is, at least we want to make sure that the economics, the pricing also becomes more digital friendly, right? It becomes more transactional focused. So that one only pays for things that they use rather than just saying, okay, now you’ve just bought data capacity worth hundred thousand dollars and you don’t know how to fill that, right? And now you’re just unhappy that, oh, somebody just took you for, by misstating things. And so we actually move away from that and just saying let’s do more simple transaction-based pricing, that’s more digital. Let just helps people get comfortable with things, right? It makes sense after awhile, once you have an idea of things throughputs, then you can always have the next level conversations.
– [Calum] Yeah, that’s a trend I’ve found really interesting that’s playing out, is this move from paying for things just as you need them and like when value is delivered. So an example, most people are probably familiar with is like Uber, where you pay based on the distance driven and the time taken and also like different surge pricing depending on demand at that time. But you don’t need, it’s all factored into, or there’s different factors that feed into this price. And then your information’s already in there, you just get charged per the value upon receiving the value and you’re getting dropped off at the location. And I think similarly, when it comes to many other spaces there is an increasing move towards just paying for capacity as you need it. This is also true of the cloud providers. So rather than buying the server, and then now you might have excess capacity, or you get overloaded and now you need to buy more servers because you don’t have enough, you can scale up or down based on what you need. And that’s one of the great things about using cloud services is paying as you need it. But I’m curious, what does that actually look like for you? So for listeners that are saying, yeah, that’s all I’m good, I understand cloud, I understand Uber, what does that mean for SAP? And I’m not sure how much you can talk about this from like the cost modeling or the pricing structure you do, but I would love to hear some more details on like what the actually it looks like.
– [Rohit] It looks, so far, I mean, I can at least generally talk about like how we think about the IoT Connect service. So like there are two components. One is the actual access to the digital, the API management capabilities, managing security, setting up like data limits because what you don’t want is devices going rogue, right? Like imagine you have an IoT connected device, deploy that device, gets hacked, and now it has been used for nefarious purposes, or you’re racking up roaming charges on that device. And so putting all of that to the software layer there’s a simple fee set up for that access. And then of course there is the data transmission, and the data transmission is very transparent pricing which is like, hey this is the pennies that you are, less than fractions of pennies that you pay on per megabit of data that is transacted. And then it’s just, depending on the actual usage, it then gets billed to the customer. So if somebody who’s then starting out with like, as said as low as like a few hundred terms they just know, that look my expenses on connecting with you, we’ll probably run into like a few thousand dollars a month, right? It’s not like it’s going into tens of thousands of dollars, right? They know that it is there and those months where they’re not doing heavy testing, or they have fine tuned the device to like only send certain heartbeat signals, then their transaction costs actually drop down to very manageable, a few hundreds of dollars, right? And that’s quite okay because it is, we see those different Applications, right? There will be certain Applications where there’ll be high throughput and certain cases where there’ll be low throughput. And I think it just works out well. And frankly, it gives customers the comfort of knowing when they scale, right? Like then they have this capability.
– [Calum] Cool, cool, yeah, so it sounds like tracking per databases. And part of my reason for asking was that I’ve also seen your per device models or per location, and it can be tough from a customer perspective, ’cause they wanna look ahead and say, okay, yeah, like what is this gonna look like as we scale so they can do that cost modeling. I think it’s important to tie as much as possible, the pricing to the value. So in this case, if a lot of the value is being generated from the data passing through then one that’s something you can track. And two, that’s the value. So if they’re passing more data, then they’re probably getting more value. So then you’ll be willing–
– Yeah.
– To wait for that.
– [Rohit] Exactly, that’s actually a really interesting way to look at it, yeah, I mean, I only thought of data to put as things but then yeah, that’s actually a very interesting way of looking at it, right? Because there’s a reason why you’re passing large amounts of data, it be because you’re probably getting more value and therefore the transmission charges are appropriate. So yeah, that’s a very interesting point. And as you said, you wanted therefore limit the times where it’s not providing value, like if it’s nefarious or if there’s a rogue–
– Yeah.
– Device that’s just sending a ton of data, which is really important. So either having some sort of rate limiting, or just a flag of, hey, this device is acting really strangely and sending a ton of data, so that you can look in–
– Exactly.
– Yeah.
– [Ryan] One of the things I’d be curious to hear about since we’ve been talking about kind of the scaling of of IoT solutions is from your experience, or what would you say is one of the biggest hurdles to IoT solutions being able to scale? We talk a lot about connectivity. We talk a lot about hardware, people seem to be very focused on the hardware and forget the connectivity aspect, but at what stage of planning, or I guess, is there a single component of a typical IoT solution that you’ve kind of run into and has sort of been neither the bottleneck, or the hurdle to allow solutions to scale? ‘Cause I know from our experience, one of the biggest areas is kinda in the hardware and it’s for a variety of reasons, but I’d be curious to kind of hear a little bit more from the experience of the projects you worked on. If there’s been a common element, or component of an IoT solution that’s really caused headaches and maybe a solution you thought was a great solution, a great idea, but it wasn’t able to scale because of X.
– [Rohit] So let me answer that question by actually sort of taking a different tact, which is that there are certain industries where I think this works very well, because I think they have tuned their value realization capabilities much more already, right? Like, so when it comes to logistics, when it comes to transportation, automotive manufacturing, heavy equipment. I think those industries have already geared themselves to this whole, the RFID, the machine to machine communication, things the processes are fine tuned to it. So I think they make great usage of those scenarios where connectivity and IoT works very well. Other industries as they are still playing with things, I think what happens is that, in my view sometimes in a POC it works well because you may have done some custom fittings in the end point itself attached to a sensor, et cetera, and then it works great, right? The challenge then comes is that like how do you go ahead and retrofit all of those existing equipment out in the field, right? Like with the sensors that can then have this communication capabilities. And I think that is something that is often frustrating for the customers to actually scale, right? And I think, those are the hardest ones where we say that, look the value seems it is there, it is a pattern that yes there will be things that we can realize, the customer sees the things, but just the whole element of retrofitting things and not having the luxury to start greenfield is the one that comes most frustrating.
– [Ryan] Yeah, that makes a lot of sense. I think on either one or two episodes ago, we we’re talking to somebody about a similar situation and obviously an ideal world being able to have everything built from a greenfield angle would be great, but the reality is that’s not always the case. So yeah, and actually I wanna touch on that because this was, I think, a misconception of mine coming into the IoT space was thinking that, well, hey, we have all these different technologies, like this should be easy in some sense. And I think the challenge isn’t that we don’t have a lot of the technology in place, I mean, I know much more to come advancements to be made, but in a lot of cases doing very simple things just like tracking assets or whatever it may be, we have the fundamental technology, but the challenge is in retrofitting and in having it fit into existing systems or processes. I think, we worked with companies that are still using like the AS/400 IBM system, which is from, I mean, like the ’80s, which was shocking to encounter. But it was also a really important realization for me personally, like, oh wow, we aren’t dealing with a greenfield as you said, so we not only need to consider is this technology possible in a vacuum, but how does it integrate into existing business processes, or existing business systems, so that it can provide value without having to change the entire organization from scratch?
– [Rohit] Yeah, I mean, and I think, that is the very irrelevant, but the hardest problem to solve. I mean, I’ll draw an analogy from something that is often being talked about and most recently at this Mobile World Congress in Barcelona and that was 5G, right? I think one of the biggest things that we will get with 5G is expanded connectivity options, bandwidth improvements so that latency gets reduced. And I was reading this somewhere that if you think about the human eye, the human eye has, we have a latency of about 13 milliseconds which means that even that occur less than 13 milliseconds, we can distinguish between them, like to us, they’re all real time. Whereas with 5G, you will actually get like latency about 10 to 11 milliseconds which means all of those science-fiction things of like the doctors sitting remotely in one part of the world moving a robot and a patient in another part of the world getting in and the robot and moving and doing the surgery could in principle be possible with this, right? The question though, is, is all the retrofits there right? And the capability and actually the training of the person, that the people element of these things, right? Like, so just because now we have low latency because a 5G doesn’t mean that it’ll all be done, right? It’s just possible now, so I think, oftentimes we get excited too much of like, oh, this is here and now we should start seeing this in the next six months to a year. And then often forgetting that, hey, there’s a whole bunch of things that would have to go through like regulatory approvals, the training, process adjustments, retrofitting, et cetera. So I think those are the elements that I would encourage like people to always keep in mind, right? Like don’t lose excitement about the future, but then also being in lens of pragmatism to it.
– [Calum] Yeah, and I think that’s what we see driving hype cycles of things where you see these initial possibilities with technologies, whether it’s 5G or–
– Blockchain.
– Yeah, blockchain, or machine learning. As the people get excited–
– Great.
– [Calum] Their excitement, isn’t misplaced, like these will be transformative technologies, but it won’t be instantaneous. And so that’s why we see this big spike up in interest and then it kind of dies off, but then the people who stay around and continue working on it are the ones that actually get it woven into what already exists, not necessarily like replacing totally because it doesn’t work like that. We need to build from what already exists, the infrastructure that’s here, but over time it can be transformative. And I think that’s why the term digital transformation is that transformation, like a transformation isn’t instantaneous, it’s not digital replacement, but digital transformation of organizations because they need to take what they already have and then transform into a digital organization piece by piece.
– [Rohit] Yeah, I mean, it’s much like any change management exercise, right? Like one has to look at like in their organizations and see where they are on the ready, willing, and able metrics, right? And just because we think digital transformation is the right thing to do, it doesn’t mean that it’s right for everybody immediately, right? It takes that, one has to keep that element in mind.
– [Ryan] I’d love for you to kinda touch on, now we brought up 5G, which I think at this point we should talk a little bit more about it, since you just mentioned you were at Mobile World Congress Barcelona, We were at a CES back in January and 5G was obviously all the hype. But for a lot of listeners out there, I mean that’s all they’re hearing is kind of the hype, or at least they’re singing it headlines, but they don’t have much context to kind of what it possibly could do, or the potential it has in regard to how it could transform the world of IoT. In your opinion, what do you see as the biggest benefits of 5G? Like what are you looking forward to the most in its connection and potential in the way it could affect the IoT space?
– [Rohit] Yeah, I mean, I think for me, the longterm vision of 5G is, and I now I’ll just throw out a full little acronym, it is massive machine type communications, MMTC and like providing ultra-reliable low-latency communications. I think, these are the things that what 5G will bring, right? Like now having turned those complex words out there, here’s how I perceive this, right? Today, if you look at nature, right? Look at human beings and how the human eye is. The human eye is actually a fairly dumb IoT sensor. It’s essentially a camera that picks up images and then keeps transmitting all of it back to the central nervous system, that’s the brain where these are comprehended and we take actions on it, right? This is an example of a smart core and then a relatively simple edge. On the other hand, a frog’s eye is actually an eye that like, it does a lot of local processing. It transmits only limited things back to the brain, right? So I think that’s the advantage of like where the intelligence sits at the edges and then less at the center. And I think what 5G brings is then the ability to transform things between, in these two analogies, in certain cases, 5G will actually help make certain very complex edges simple. So if you think of autonomous cars today, an autonomous car today is actually a very self-contained device because the latency and connect with the, provided today is not enough for the car to function, it carries everything with it like in all the LIDAR, all the other details. The only communication that it is doing to the center is for like basic GPS and maps and a few connectivity things because it just cannot rely on the available transmission capability. With 5G that has a big potential to get transformed completely, right? Which means that each autonomous car can then be more distributed. They can then have, because of the latency issues are not there, they connect with the options have improved, perhaps we will see a bigger, wider adoption of those. So that’s one example of like how 5G drives that capability. The other one is the earlier one that I was talking about in in the field of medicine, in the field of remote surgery, robotic surgeries, right? Like now latency is now used which means that a lot of those things start becoming capable and more effective. So I think those are the kind of transformations that I think 5G has the capability to make happen which we don’t see today.
– [Calum] So do you think there’s, I may have misheard, but are you saying that there would also be examples of things becoming more like the frog’s eye? ‘Cause I, yeah, go ahead.
– [Rohit] Yeah, now, I think that there will be an example of like where the edges will actually become much more smarter, right? Like, so where, and today what certain scenarios, because the edge can not become that smart. And this is where it’s got less to do with 5G, but it has got to do all with the increase in compute capabilities, increase in AI and distributed deployment, that the edges will be able to like sense, respond, manage things locally. And then only transferred back a few things back to the center, right? And therefore they reduced, they’re connected with the dilemma of like, because today a lot of those things cannot be done just because they say, look, I don’t have good connectivity options and my edge is not smart enough, right? Like now with this, there will be certain situations, and especially if you think of remote oil wells, right? All oil rigs and others, right? There you don’t have sometimes the option of providing big connectivity pipes back to the center. And what you need is really smart edges where they can sense and respond and react things. And then only do a limited callback, right, into the center. And I think that’s where it will become a little bit more like the frog’s eye.
– [Calum] Yeah, so to expand on that, the way I think about it is the reasons in favor of the human eye approach, where you have the sensors are dumb, they don’t do a lot of processing on the edge and just pass data back into some center, which does the majority of the intelligence work. The advantage of doing that is that you can not have to put that intelligence in each sensor that’s on the edge, and that can be make things less expensive from a hardware standpoint. So you don’t need all these smart things, but you can have the smartness back in that centralized location and a way of making this real for people listening is like, just think of your phone. So the fact that you have access to all of these different apps on your phone is because most of the intelligence is being performed in the cloud. So your phone just needs a good connection and you have a window or a conduit into extremely powerful computers and systems that could never fit onto your phone, but don’t have to, because you can just access it remotely. So those are reasons in favor of having intelligence towards the center, but to your point, in some cases, you might not have that conduit where if the connection from your phone to the cloud was just in the kilobytes per second, you wouldn’t be able to watch videos, or do many of the things we like to do today. So if it’s an oil rig out in the ocean, you just don’t have that connectivity, or you probably want intelligence there. So you can continue to do these things that require intelligence given that you don’t have that connection. But it’s interesting bringing up autonomous vehicles and hospitals, because much of the reason, another reason you might want to have intelligence on the edge is in cases where latency in mission, critical situations could be an impact. So for the autonomous car, if you are driving and you are about to crash, you don’t wanna wait on your car to have to send a message up saying, hey what do I do? And then listen for some algorithms on that far away to then come back and tell it. In those milliseconds you might’ve already crashed. And similarly in a medical situation, or even in like a military context you might want to have the intelligence as close as possible because every little millisecond might count. But something I wanna follow up on related to 5G, and pardon my ignorance here, but when you mentioned the, you 10 to 11 milliseconds, is that just from the cellular towers to devices? ‘Cause I would imagine that there would also be some latency from like a processing standpoint. So to send it up, get the messages processed, perform some, whether it’s analytics, or some sort of algorithm on the data being passed up and then send the command, or the information back to the device.
– [Rohit] So now you are also pushing my limits of know-how on that. And my understanding on this is that like the 5G will be more distributed. There would not be like no more fuel cell towers driving that, it’s rather a distributed network of mini towers, so to say, like the mini towers would be across and then that creates that grid. And that’s why you get a better response time and latency. Unfortunately, that’s where the limits of my know-how and technical depths end on this. I wish I had more, but yeah that’s probably why the latency is improved. The question though is still, even though just because it is improved, doesn’t mean that it will all be implemented immediately, right? Like there will be still that element of experimentation. And my guess is that as always first we will start to see this emerge in more AR, VR and gaming experiences before we actually get to the real surgeries, right? Because it’s quite okay to test and get the kinks out in an area, a gaming environment, because you still need high connect with being low latency to make it as real and immersive as it can be. But I think we will be more forgiving if we don’t get the game right as opposed to getting the surgery right.
– [Ryan] Some avid gamers may disagree with that.
– [Rohit] Yes, I agree, I agree. I was thinking about that as I was saying that, yes.
– [calum] Yeah, I think there’s some for him, it is life and death. The results or something.
– Yeah.
– I think, this is probably a good opportunity for us to transition into the AskIoT questions for the show. So we can bring in Shannon, who has I think three questions to ask us.
– [Shannon] Hi, Rohit how are you?
– [Rohit] I’m doing well, thanks. Okay, my thought. So I was doing some research on you and I know that you deal with omnichannel communication, can you explain to us what that is and how it impacts IoT?
– [Rohit] Yeah, so the omnichannel communication is, as I was talking about SAP Digital Interconnect and our approach of intelligently connecting everyone, everything everywhere. So the omnichannel experiences start from connecting everyone, right? And what this means is that helping businesses reach their end customers in places or in channels where the customers are present as opposed to forcing the customers to come to them, right? Like, so our approach is that, hey if the customer is present on SMS, then we help you reach there. If the customer is present on one of the social apps then we help you reach there, or email, right? And the fact that all of these, it’s just because now there are social channels, it doesn’t mean that the SMS is dead, right? Or not relevant. There is still value for that channel because if you think about it, there is only one app that is guaranteed to work across all mobile platforms and across the globe as long as you have cell connectivity and that is an SMS, right? So there is still something to be said about that channel. And I think it’s that same philosophy that we bring into IoT, which is, and look, there will be multiple forms and formats of IoT connectivity. And we want to bring all of those to our customers and hide that complexity of managing across between them. Does that answer the question?
– [Shannon] Yeah, definitely, I think that was great. How do you foresee consumer behavior impacting IoT in particular with this omnichannel communication and bringing the consumer to the product as opposed to the other way around?
– [Rohit] I think, it’s, I mean, if you think about it, it is occurring already and it’s sometimes just a matter of people changing their approach to things, right? Like, I mean, early in the show, we were talking about and saying, well, think about your mobile phones, well, that’s an IoT device. And think about your car, that’s an IoT device, right? Think about now, I just upgraded my home appliances and my refrigerator now actually is connected to WiFi. And then if the door is left open by my teenage son, I do get a text so that I can then harass him and saying go close the door, right? That’s an IoT device and connected. So I think more and more we will see these coming into the smart home. And that’s where the consumers will then start driving the need back to the manufacturers and others, right? Like, because when they will start seeing the end user benefits, right? It’ll also bring up an interesting challenge which is that more and more, and customers will also start thinking about like what information and what data are they sharing and will require the manufacturers be upfront to them about that as well.
– [Shannon] Very good, so one last question. What is the role of data in IoT, just a high level understanding for our listeners who may not know a lot about IoT?
– [Rohit] Data is the, it’s the foundation of providing any intelligence that we build on in this world of internet of things like things, it’s the foundation on which we then build information to act on it, right? And I think just by saying, generate data and collect it, doesn’t do good to anybody else, right? But on the other hand being able to collect this information and bring it in a meaningful format, I think that is the real power. I mean, just to put it in context, I think there was a study we just referenced by the end of 2018, the data created by IoT devices will be 403 trillion gigabytes. And I think right now it may be the stage where we are actually collecting more data rather than acting on it. But I think over time we will see this improving And in fact, there are some estimates which show that very soon machine generated data will actually take over today what we refer to as like, voice or human generated data what we may already have. So I think it is actually quite relevant in the foundation of IoT.
– [Calum] Yeah, if I may also add thoughts to that, so going back to your comment about the eyeballs, I often think of the human body as analogy for some things in IoT. And so just imagine if you were to not have any of your senses, so you can see, you can hear, you can taste, you can touch. So there really be no way of getting in data from the world. It make it very hard to learn and make it very hard to act and know what your actions are doing in the environment. I mean, without that input from the outside, very little would be possible. And so when it comes to IoT, we’re now getting inputs from things in the real world. And I often think of the internet in a few different waves. So in the first wave, there were machines connected directly to other machines. I mean, that is the foundation of the internet, is computers that are now networked with each other. The second wave, I think driven very largely by mobile is lots of data coming from people entering it. So whether that’s passively, like people how they interact with Facebook, what they’re searching on Google, they’re putting inputs in. So those inputs could be liking things, It could be searching for something, but the data is being generated by humans. And so much of it has been created by humans. And so now this third wave, which is really summed up in the internet of things is now getting the rest of the environment, not just machines talking to each other and humans enter data machines, but now drawing data from our environments, whether that’s temperature, whether that’s like power or location of things in the real world, whatever it is, getting that, and the more data you have, it opens the potential to make more intelligent decisions, to automatically perform actions, but not necessarily so. And I think to some of your point, it has to actually be made useful. And so many companies are just collecting data for the sake of it, because it can be valuable, but if you don’t have a plan for it, one, that could be problematic from a privacy standpoint if you’re collecting data that you don’t actually need on people, but even aside from people, if it’s things that you’re collecting data on, but don’t necessarily need may or may not be valuable. So not only thinking through yes, like data is extremely valuable to collect, because it enables more intelligent decisions, because it enables automatically performing actions without needing a human in the loop. But also–
– Yeah.
– [Calum] Think through what is the value you’re trying to create and not just collect data for the sake of collecting data.
– [Rohit] Yeah I mean, and that’s very well put right there. I think for us at SAP, human centrism is the core principle of our vision, right? Like as we talk of intelligent enterprise. And actually it aligns with our mission to help the world run better, right? And improve people’s lives because in the end, that’s what we should be focused on eventually, right? Like otherwise what’s the point, right? Like, so I think you’re quite right, that human centrism is very important in these discussions.
– [Ryan] Yeah, I mean, at the end of the day without the data IoT is, and that’s at the foundation of what IoT is, right? Like without it, it’s not gonna matter, and they won’t be doing it ’cause the data is what is allowing people to make better decisions and kind of improve the processes and whatever it is that they’re working on. So any other questions, Calum, we may also follow with? All right, cool, I think, that’s pretty good place for us to stop. I like to wrap up the show, right? With you kind of seeing if there’s anything you wanna leave our audience with, anything you kinda wanna mention before we get off something maybe we talked about already, or something else that we didn’t get a chance to chat about that you think is important?
– [Rohit] Well, I mean, I’ll just quickly recap and you’re saying that look, we at SAP Digital Interconnect with our service of SAP IoT Connect 365, we see a massive opportunity to help our customers and our listeners to take and complete their digital journey, and not just to innovate, but then also to become truly scaled digital businesses. And I welcome them to connect with us on our Twitter handle @SAPInterconnect and follow us there. And we often share some of our learnings and we will be very, very excited to hear back from them as well.
– [Ryan] Awesome, that kinda answered my next question. So yeah, otherwise, I appreciate you being on with us today. I think this is great. It’s ton of value that was communicated, a lot of good questions, a lot of good answers. So thank you for taking the time to be on the show with us today, and hopefully we’ll be able to have you back sometime soon.
– [Rohit] Awesome, thank you guys.
– [Ryan] Hi everyone, thanks again for joining us this week on the IoT For All podcast. I hope you enjoyed this episode and if you did, please leave us a rating or review and be sure to subscribe to our podcast on whichever platform you’re listening to us on. Also, if you have a guest you’d like to see on the show, please drop us a note @ryaniotforall.com and we’ll do everything we can to get them as a future guest. Other than that, thanks again for listening and we’ll see you next time.
Industry News
