Transcript.

Erik: Welcome to the Industrial IoT Spotlight, your number one spot for insight from industrial IoT thought leaders who are transforming businesses today with your host, Erik Walenza.

Welcome back to the Industrial IoT Spotlight podcast. I'm your host, Erik Walenza, CEO of IoT ONE, the consultancy that specializes in supporting digital transformation of operations and businesses. Our guest today is Frederic Desbiens, Program Manager of IoT and edge computing at the Eclipse Foundation.

The Eclipse Foundation provides a mature, scalable, and business friendly environment for over 375 open source projects, including frameworks for edge and cloud applications, IoT and AI. In this talk, we discussed the Eclipse Foundation’s recently published IoT and edge commercial adoption survey, which is released every two years; we explored edge computing adoption trends, future investment plans, common implementation challenges, and changing perspectives on the dynamics between edge and cloud.

If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Finally, if you have an IoT research strategy or training initiative that you would like to discuss, you can email me directly at erik.walenza@IoTone.com. Thank you.

Frederic, thank you so much for joining us today.

Frederic: Thank you so much for having me today.

Erik: So we have a very specific topic today we have a report on edge computing, and IoT to discuss which I'm really looking forward to discussing with you. But before we go there, Frederic, I'd love to learn a little bit around how you ended up with the Eclipse Foundation. So I know you've worked with some of the larger players in the field with Cisco, with Oracle, you've also worked with Pivotal Software. How is it that you found yourself now with the Eclipse Foundation?

Frederic: So essentially, what happened is that for a long time I've been a software developer and architect and then I got into product management and at that time, I joined Oracle in 2012. And I was involved in a team that was dealing with development tools and mobile technologies, mobile Backend as a Service.

After that, I went to Cisco, and then I got to start working intensively with IoT. I had a short stint at Pivotal as one of the program managers or the product managers for Cloud Foundry Bosh, which is a component of Cloud Foundry that orchestrates virtual machines. And at the time, I was simply, not necessarily looking for a new job or anything, but we wanted to get out of the Toronto area which is very expensive. And I was simply on the prowl for good opportunities, and I stumbled upon this offer from the Eclipse Foundation, and I joined in February 2019. I'm lucky to have this job. That's certainly the best job I had in my career.

Erik: So it says here on your LinkedIn profile, you're managing three different clips working group that, in total, encompass 50 open source projects. Maybe this is a good introduction to the Eclipse Foundation, if you just explain a little bit, what does that mean that you have these three working groups and then 50 open source projects underneath them?

Frederic: And before we get there, let me provide a bit more context about the Foundation as a whole. Because the three working groups I take care of are 3 out of 17, so we do many more things than just what I do, obviously. The Eclipse Foundation, even today is still associated with the classical desktop ID, which bears the same name, Eclipse ID. And Eclipse ID was born in 2001 an open source by IBM. In 2004, the Foundation was created in order to be the vendor neutral steward of the Eclipse IDE and the Eclipse rich client platform.

And from there, for a long period of time, we were doing mostly development tools. But in 2011, we started getting into more technology areas and the first one of those was IoT. So Eclipse IoT under a different name was created in November 2011. So we'll be celebrating the 10th anniversary later this year. So, overall, the Foundation as close to 400 open source projects. And in the IoT and edge domain, we have roughly speaking 50 Plus projects.

So the three working groups, I'm taking care of our Eclipse, IoT, Edge Native, and Spark Plug. So Eclipse IoT is the largest one, the oldest one as well. And essentially, the point of Eclipse IoT is to provide building blocks to build IoT solutions. And many of them are used on a daily basis in commercial IoT solutions.

In Eclipse Edge Native, we've got essentially three projects there. Two of them are full blown edge computing platforms, so that's Eclipse IO Fog and Eclipse Fog OS. And then the third one is Eclipse Zeno, which is a next generation pubsub protocol that essentially integrates the notions of geographically distributed storage as well as evolves and queries and things like that built in the protocol. And the last one is Spark Plug, which is an offshoot of Eclipse IoT. And Spark Plug is an open source specification and protocol that go on the top of MQTT to make sure that MQTT based solutions can be interoperable out of the box.

So overall, 50 plus projects, three working groups, and my role there is essentially two things. First, is just to keep an eye on our open source projects and ensure they follow the Eclipse development process, that they are responsive to their community. And then the second part is literally evangelizing the technology. So I'm always on the prowl for occasions to talk about the technology, find new adopters for it, and eventually, obviously, drive membership conversations with the people I meet.

Erik: You're obviously doing a little bit of evangelizing today, but this is also a very important role. I think open source has had tremendous growth, not just in IoT, but across the board over the past several years. And it's always interesting for me to understand also the little bit of the business model behind the organizations that are actually supporting open source. What does that look like for Eclipse? Is it a nonprofit? Is it a corporate? How is it that you sustain the activities for developing this tremendous amount of technology?

Frederic: So in our case, the way the Foundation operates is that we are vendor-neutral, member-driven, member-founded nonprofit. And what that means, essentially, is that our members pay annual fees that help us support the staff, and in turn, the staff is just there to deliver services to our open source projects and obviously, to support the working groups.

The one interesting thing is that to have an open source project at Eclipse, you don't need to be a member. Tomorrow morning, you have code that you want to open source, you can start an Eclipse project without paying anything to the Foundation. And we have a number of projects like that, that are driven by non-members. So we see ourselves really as the stewards of those projects, but ultimately, it's up to the project teams to build those.

So as a program manager, unfortunately, I don't have the time to contribute code or anything. And our staff is really focused on keeping an eye on the community and fostering its growth. But ultimately, it's the project leads and the committers and the contributors to the projects that drive, not only the technology direction, but that invest the hours to build those great open source projects.

A good example of that we've got in Eclipse IoT is what BOSH is doing. They are in automotive. They are in industrial automation, and many other sectors. And what they did essentially, is that around 2015, they open sourced a number of technologies that were proprietary to them and those became Eclipse open source projects. Well, most of those projects are still around. For example, Eclipse Hog Bit, which is a platform to distribute over the air updates to devices; Eclipse Hono, which is a high throughput message routing platform; Eclipse ditto, which is a fantastic digital twin platform, etc. And those open source projects are the pillars of the commercial BOSH IoT Suite.

So essentially, BOSH is paying close to 60 developers to work on a daily basis on those projects. And obviously, over time, those projects attracted contributors that are not BOSH employees and people that add up to the technologies and that started to contribute to the project. And then on the commercial side, what BOSH is doing is that they take this open source code and offer this software as a service in the cloud, AWS and you can use BOSH IoT Suite.

And the only thing difference between what we have in open source and what BOSH is selling as a subscription on the cloud is essentially that we don't have the shiny UI. But all the base code for the projects I mentioned, it's the same. BOSH is literally building in the open, and then put that in the cloud and then they have their appropriately UI and some value added features for commercial customers only. As you said earlier, open source is really trending right now.

And the reason why is essentially that overtime, anything becomes a commodity and it doesn't make sense to build something which is a commodity. And in some cases, if you think about the Linux operating system, it's a very complex commodity and operating system. And so building that alone in your own little corner of the internet doesn't make sense.

So that's why it's so important that people get together and build in the open things that are common to everyone, and then innovate at the top, innovate in the commercial space, using a robust and vendor-neutral open source Foundation for their projects.

Erik: You have a very diverse membership group here. I mean, you have many of the usual suspects, but also, Daimler, Fujitsu, Red Hat, you could kind of imagine would be a big player here, Airbus. But a good mix of more pure tech companies, but also, what I would consider to be end users, you have Goldman Sachs as a member, which is a bit interesting, but really quite a diverse member base.

So that's a great pool of talent to draw from, and I guess not just developer talent, but also, I think one of the critical things here is understanding of what the end users actually require. I suppose you have different roles for different players in terms of technical and business functions?

Frederic: Yes, absolutely. And you're putting your finger on something really important that makes us special in the market, in the sense they are alter open source foundations in the space, but we proud ourselves to be the most commercial friendly of them in the sense that we are not shy about the commercial successes and commercial adoption for our technologies. And the alter is that yes, our membership is diversified. It's not just IT companies speaking to each other.

In the IoT space is especially important because you have operational technology people and the more traditional information technology people, and we have people on both sides of that divide, which means that our projects really address real operational technology, challenges, and at the same time still keep an eye on current IT trends and things like that. But overall, when you consider our membership, the one thing that we do very well is that we are not a pay-to-play organization in the sense that, yes, we have membership fees that our members pay, but those fees are essentially scaled according to the annual revenue of the organization.

So if you are a small startup with an incredibly exciting innovative technology, and you go to some mature comparable organizations, what would happen is that they have very high fees for members that would be at a strategic level inside the specific consortium. And that means $50,000 or 50,000 euros and plus to have the maximum level of membership with the voting rights and everything. And to us, that's not a good model, because you are implicitly focusing on the big players that have the means to pay that kind of money. And probably for the larger players in the market, this is a rounding errors on their financial statement. But for a startup, it's a huge chunk of money.

So in our case, you can be a strategic member of the Foundation as a startup, you can be a strategic member of any of our working groups as a startup, and you will have to pay annual fees, yes, but they are scaled according to your capacity to pay. And this really makes us different because then it means that even startups can be involved in the governance of our working groups, have a voice to the table, and shape our vision and strategy. And that really what is making us unique in the market.

Erik: It's really a powerful ecosystem that you put together. I'm really interested in the topic in particular, because as you said, Eclipse has a strong also in addition to having a technical base, also a very strong commercial base. In this particular report is the 2021 IoT and edge commercial adoption survey. And this is, fortunately, a positive trend, but also a very important trend that we're starting to move beyond discussing pilot projects towards really practical deployments at scale. Before we get into the results here, can you just share a little bit about who was participating in the survey?

Frederic: Yes, absolutely. As you mentioned, we are discussing our commercial IoT and and edge commercial adoption survey, and in this case, this is the second edition of that particular survey. But since 2015, we've been running an IoT developer survey as well. And so now we are on a cadence where in the first half of the year, we have the commercial adoption one. And then next fall, we'll have yet another edition of the developer survey, which is more technical.

So in this case, for the IoT and edge commercial adoption survey, the goal was really to better understand the industry landscape for IoT and edge and really understand the challenges faced by organizations that are currently deploying using commercial IoT and edge solutions. So we conducted the survey from January of this year till mid-March, roughly speaking, and around the 300 unique individuals participated. Demographics for those individuals, there's a number of executives in there, then software architects, developers, pinch off product managers as well.

And really, we were focused on people there that really have their two ends into actual projects. And that's what makes the survey interesting because it really highlights the real challenges that they've met in working on those projects.

Erik: And the survey here is divided into three major sections, which is IoT and edge adoption, IoT and edge investment trends, and here we're looking at investment as investment in solutions, not venture capital, or this type of investment, and then lastly, strategies and challenges. So maybe we can walk through in that order starting with IoT, and edge adoption trends. But what were the things that really jumped out at you around the shift from 2019-2021?

Frederic: The first thing that we noticed is that really adoption for IoT and edge technologies accelerated even with the pandemic. Personally, I was expecting when we started to see lock downs right and left in the word last year, or roughly in March, and April, that investment in IoT and edge would decrease, that we would be in for a tough couple of years. But no, people not only continued to work on IoT and edge projects, but they tell us that they plan to even increase their investment and the number of projects that we'll be working on in the next 12-24 months.

So specifically, 47% of respondents in our survey said that currently they are deploying IoT solutions, and an additional 39% are planning to deploy in the next 12-24 months. And in the case of edge computing, 54% of organizations are either using or planning to use edge computing within 12 months, and another 30% have plans to do so in the next 12-24 months. So for me, it's really exciting because, not only we have new players in the market, but the current players are really thinking about scaling up their involvement and their investment, which means that IoT and edge are really delivering actual value to the people right now.

Erik: It was interesting for me to see where IoT and edge solutions are being deployed. So, of course, you have the usual suspects, predictive maintenance, manufacturing, and logistics and supply chains, which makes a great deal of sense. But also, I think number one among the different buckets here, product and service development also very high in in research. So with about 14% of deployments in research, 8% in sales and marketing, and another 6% in customer support.

Because often when we think about IoT and edge, we think about a manufacturer and operator deploying this in a facility which I think intuitively generally makes sense for people. On the business side, where are you seeing the major adoptions in these topics of sales and marketing, customer support, product and service or R&D?

Frederic: There are so many interesting use cases in there. It will be hard to pick just one. But when you think about, let's say marketing, for example, obviously, retail is an especially good target for IoT and edge deployments. Because essentially, this is all about delivering tailored customer experience for people when they go physically in a store, or even when they deal remotely with your organization. The value there is tremendous.

Obviously, one thing to consider is that people when they go to the store, they want a tailored experience, but then they don't want you to track every little gesture they do in great detail. And they don't want especially the store to be able to trace back your specific behavior as a customer to the data they are gathering. I saw really innovative ways to achieve both without compromising privacy, which is an important concern for both organizations and consumers.

So a good example of that was a specific retailer was deploying an IoT enabled tracking solution in their stores, and instead of putting cameras all over the place and recording the faces of people, what they were doing is that the deployed cameras close to the ground and they were doing edge-enabled AI recognition on the cameras. So those were fairly powerful cameras with the built-in processing power. And what they were doing is that they were tracking the individual pair of shoes as they were going through the store. And then this obviously provided data about what were the most effective setups in the store, where to where to people paying attention to a specific display, and things like that, and spending some time inside the store.

But the nice thing is that by tracking just the tubes, you were able to distinguish individuals but without tying this to their actual identity. And to me, innovative solutions like that are a great showcase of what IoT and edge can deliver. Yes, there are privacy concerns. But when the suppliers of solutions, and when the actual customers pay attention to those aspects, it's possible to have interesting deployments and actual value for everyone involved.

Erik: Yeah, this is an ongoing learning experience for everybody to figure out how to realize the value while minimizing the privacy concerns here. Maybe you can let me know if this is an interesting trend in one direction or not. You have a data point here on who's actually deploying these and 35% of the solutions are being deployed by the companies themselves. So they're basically sourcing technologies, and then developing their internal teams.

There's been a lot of development in low code with, of course, the intention of allowing companies in particular non-technical people to take a greater role in building applications. But here, we're just looking at the 2021 data points, so we don't have 2019. In your experience, are we seeing a shift towards a larger or smaller percentage of companies that are deploying IoT themselves without system integrators?

Frederic: Yes. And open source is certainly a factor in that growth. Obviously, the tricky part about the IoT market is to get started, to pick the right sensors, to pick the right technologies and then articulate them. But given the growing maturity of open source solutions, it's become more easier and easier for people to get started and built from the ground up on their own relying on their knowledge of their own industrial processes and whatever legacy equipment they would have. And this is especially important in industrial automation in the sense that most organizations, they won't refresh the factory floor and rip everything in the factory every five years.

More often than not, you're trying to really leverage your capital investments on the long term, which means that you probably have on your factory floor machines that date back to the 80s and there's no compelling business reason to replace all of them at once. So what I saw a lot is  organizations shopping for fairly affordable sensors in the marketplace, and then using their own knowledge of their industrial processes retrofit their existing setups to get a whole slew of data that they weren't gathering before.

And this is used, for example, for predictive maintenance when you have those older machines that need a bit of pampering, but also simply to drive innovation about industrial processes, you're making better decisions if you get more data. And now given the wide array of sensors in the market that you can procure for an affordable price, it's easier for organizations to just, let's say, take an MQTT broker and start deploying sensors, gather data, and then the projects over time grow organically in this fashion, and the enabler there is to increasing maturity of open source solutions.

In any case, even if you go to commercial route, and you deal with a systems integrator in order to help you out, in IoT and Edge, there's no one stop shop, you will have to deal with multiple suppliers in any case, because maybe you need specific sensors that are specific to a segment of the market, maybe your preferred integrator is not familiar with that segment of the market so they will have to deal with otter subcontractors that are more specialized, etc. So the fact that IoT and Edge is essentially an erogenous market means that it's a bit easier for organizations to take more of the job on their own. So to me, your insight is right, we see more people doing things on their own, and using relying on their operational technology teams in order to do the work.

Erik: I mean, you're IoT ONE, my firm, we're a consultancy that focuses on IoT. But people often ask, when we're working on a project, hey, can you deploy this? Of course, we cannot deploy this because it's such a long tail of solutions that you really need to find the right specialists and then you need to have the business understanding which companies have and even the larger firms finding a company that has competence to deploy any random use case that you might pick out of a box, you're just not going to find the best practice provider there. You really need to do a proper search and find the right fit for your particular situation.

So it's a very interesting market because as opposed to CRM, or ERP, where you have just a handful of big players, and you can kind of expect people to be able to work with those big solutions. Here, it's really a case of finding the right tool for your need in a very, very diverse market.

Frederic: Yeah, you're right there's no cookie cutter solution in industrial automation, especially because the factory processes are so specific to an organization. And this is both a challenge, but it's an opportunity for players in all sizes in all corners of the market.

Erik: There's another point here as we move on to the investment topic around the range of investments. So if we look at 2021, we have about 40% that are investing above 100,000 in the fiscal year; out of that, 13% investing more than 1 million. How do you see this trending over time, the range of investment volumes?

Frederic: That's the one thing that really surprised me in this year's edition of the survey in the sense that with the pandemic, I was expecting people to say, hey, we will maybe maintain our investments, but maybe we'll cut them, but no. Essentially, what we saw in the survey is that most organizations will increase their investment for the next fiscal year. So, specifically, when I when I look at the data point, 62% of organizations projected to spend over 100k on IoT and Edge solutions into current here.

In fact, this probably comes partly from the fact that now IoT and Edge projects are not driven exclusively by operational technology, people in the factory, or maybe by a few crazy members of the IT team experimenting with sensors and Raspberry Pi's in a dark corner of the warehouse. But the survey indicated that 35% of organizations see their IoT and Edge decisions being made by the executives. And this was 18% in 2019 survey, which means that's really the C suite is starting to take IoT and Edge very, very seriously. And in some cases, some organizations will have a mandate coming from the top that they should look at how they can leverage the technology.

So to me, there's a direct correlation between this increasing involvement of the C suite and the clear trends that we saw in the survey towards higher investment for the next fiscal year. And this is good news if you are a consultancy, or solutions provider in the market. So not only it's pretty profitable right now, but there will be many more opportunities for years to come. So if you're not in the market, now it's a good time to jump.

Erik: And it's a very clear indication that this is becoming strategic growth topic for companies, where in the past, it was either a tactical, like, operational efficiencies around the edge, or it was more run some pilot projects and see if anything sticks. But now companies are saying, okay, we actually have to develop a longer term strategy around how we integrate these technologies into our operations. That's really a very important shift, not just a data point, but it really will impact how the organization's staff build organizations to develop edge.

And then there's this other data point, which is collaborative efforts between IT, OT, and BU, which also increased significantly. So it means again, there's a shift from isolated IT initiatives or operational initiatives towards initiatives that are integrating the resources from these different organizations. And again, I think that reflects larger scale in an effort to make more meaningful deployments as opposed to maybe pilot deployments, which were being rolled out in the past.

Frederic: Yes, absolutely. And it's exciting to see at last that we are starting to bridge the gap between OT and IT. Although, the typical developer profile on each side of that divide remains pretty, not the same bunch, and they don't have the same mentality in the sense that an OT developer will put a premium on stability and we'll be very careful about making updates, because if you push a patch, and then all of your robots are offline for six hours, that's already a massive financial loss for the organization. Whereas in IT, we are used with DevOps, doing things in a more [inaudible 32:27] way and patching in the middle of the day, and then you just refresh the page and the battery's dead.

But to see those two aspects of organizations converge is pretty exciting to me, because it's only when we have both sides of the equation together that really we can capitalize on all the benefits and potential of IoT and Edge technologies.

Erik: And Frederic, if we move on to the strategies and challenges, this was really one of the most interesting data points and probably very excited for Eclipse. But this question around proprietary versus open source solutions for IoT and Edge software, so there was really a significant jump in companies that are purely looking at open source solutions. So from 23% to 39%, between 2019 and 2021, and then a similar decline in companies that were looking for proprietary. There was a good a good group of companies that were kind of unsure what their strategy was and that's declined. So this change in companies that are using pure open source, what do you think is behind that trend? What are the primary decision criteria that would be driving that?

Frederic: To me, you would think since open source is free, that cost is the driving factor, right, intuitively? But in fact, we asked that specific question to people to pick the number one reason that they are using open source software or looking at it if they are not already using it. And the number one reason with roughly speaking, 30% of the respondents picking that one is that it's the ability to influence or customize the code. And when you think about it, it completely makes sense in the sense IoT is all about constrained devices; you have those microcontrollers. So there's a premium on memory on program size and on power consumption, even if you think about some devices that would be battery operated in remote places.

So since it's so important to tailor the software, and even the hardware to the specific use case, to me, It completely makes sense that it's this ability to customize code that that really was the number one driver between open source adoption. Then in the survey, yeah, the cost advantage while was also there at 18%. But the number three reason was flexibility. And once again, when you consider some of the missteps in the IoT and edge market, you understand why having an open source based strategy is so important in the sense that I won't name names there, but we've seen many big players, especially in the smart home market, and what it is they consolidated they brought smaller players that had an installed base and things like that and then suddenly, they discontinued the products in order to consolidate on a lesser number of platform, but this literally left people with broken smart homes. You have all of those sensors in your walls and everything, and now the hub and software that we're relying on is discontinued and then you're stuck with a broken home.

So obviously, if you are working with components that are open source, you are not at risk of being stuck like that, because either you can maintain the code base yourself on the long roll, or hire someone to maintain it if you don't have the skills internally. So once again, when you think on the long term, and especially industrial automation, you don't have a choice but to think over the long term because the machines that you have on the factory floor are expensive. So the fact that open source is there, and the fact that you can customize, cut the features you don't need and really focus on your use case and ensure that you have hardware and software that really are tailored to your specific deployment, to me, it completely makes sense. You cannot do this in a proprietary fashion and this is why people are so interested in open source.

Erik: How do you see open source and the topic of security? So security is listed here as the number one concern for IoT and edge. And open source, I suppose on the one hand, you could say that because it's open, okay, it's possible for people to look into the code and find vulnerabilities, but also because you have a larger ecosystem that's chasing bugs and identifying fixes, you can also have a stronger solution. What would be your maybe your advice here for companies that are adopting open source solutions but have an overall high concern for security of edge devices?

Frederic: Security is an important concern for all of the open source projects that we've got. And the one thing to look for is really, when you look at open source solutions, not every open source solution is the same. If this is just a single man project on GitHub, that's not necessarily the same level of structure. And I would say that you should not necessarily have in place the proper processes to FCVs when they have [inaudible 38:00] in the code base, and things like that, which is something that we have our projects in our case at the Eclipse Foundation delivered to the community.

So look for CVEs, look for projects that obviously have frequent releases and pay attention to the topic. But in a wider perspective, the number one advantage of open source in a security perspective is the fact that when you pick an open source component, especially to implement specific security features in your project, you are not writing the code yourself. Security code is probably the trickiest sort of code that you can write. And if you just rely on your own knowledge of the basic standards and the basic processes, you will miss on a lot of potential issues and then you will be on the hook to stay abreast of the latest trends in attacks and the latest trends in the sorts of exploits that are in the wild.

So the fact is, if you focus really on using open source blocks that are well maintained, and that are not necessarily single vendor projects, but supported by a wide array of contributors, then it means you don't have to write the security code yourself, and other people are doing the job of tracking this and tracking vulnerabilities and informing the wider community when they are vulnerabilities that should be patched.

The key thing that you should do if you incorporate those open source components and rely on them is that you should be proactive about adopting the latest versions to ensure that you're not using something that is out of date and more vulnerable by definition. Because essentially, security is a moving target: hackers become more clever every day, and they will change over time, the type of exploits, the type of attacks that they will launch. And maybe as an IT professional, you don't have the time to track that. So relying on well curated open source components is especially important there.

Erik: But this is another area where we've seen a fairly signal can shift over the past two years. And the shift here is primarily in the direction of hybrid cloud, maybe it's not too surprising. Would do you think this is going to be just generally a long term trend towards more complex cloud architectures? Or are there any other trends that you see in terms of how particularly B2B companies are adopting cloud for IoT and Edge deployments?

Frederic: To me, hybrid cloud is and will be the dominant model for years and years to come. And the reason for this is exactly the same reason that's driving the growth in edge computing. In the sense that, it's great that you are able to rely on cloud infrastructure to do massive data crunching jobs and things like that so you don't have to invest in huge data centers or the like. But your proprietary data, your industrial secrets, you don't want them to leave your own servers.

So just for that reason, hybrid cloud over the long term will be the dominant model. Even organizations that say they go all in on public cloud and drop data centers, right and left, they won't send the entirety of their operational data straight to the cloud where it could be more recent breach. Some of it will be kept on operational technology infrastructure that will still be owned by the organization. And when we say hybrid cloud, it's important to define what we're talking about there. Simply having servers of your own in your own data centers is not hybrid cloud.

What we mean by either cloud is that the servers you have inside the organizations will use Kubernetes, and auto cloud like technologies in order to orchestrate the workloads. And then on the other end, some of the workloads will be in public cloud for scalability and elasticity reasons. And that model for me will be certainly the dominant model for years and years to come for sure.

Erik: Probably somehow related topic is where companies are actually using the workloads that companies are performing on edge environments. AI is number one, which Yeah, I guess, on the one hand is not so surprising, but I mean, it is still somewhat remarkable that this is so feasible today, because just a short number of years ago this really was not feasible. And all of a sudden, it's 40% of respondents saying that they're running some type of AI on the edge. Then we have control logic, which is more of a classic use case and then sensor fusion, which is also not technically simple. Anything that jumps out in terms of the specific workloads, or the trends that you see maybe moving forward around how people are using Edge computing?

Frederic: In this case, what has been really interesting in the survey is as you mentioned, the wide variety of workloads there and each of those workloads has a different profile, which means it vindicates our view that the edge computing landscape is really an heterogeneous one. Cloud is all about varied uniform resources that you drive at scale. But at the edge, you will see many, many different types of boards and chipsets and all of that. And I think a platform that that is able to tackle all of those different hardware platforms to tackle the verity of use cases is really something important when you're looking at an Edge platform.

I wasn't surprised to see AI first because I mean, when you look at it, Google Intel, many players are very affordable AI accelerators that you can plug into affordable boards in order to deliver exactly the kind of compute power you need in order to process AI. And especially in video analytics, you don't want to transmit the full HD stream at two or three gigabytes per hour directly to the cloud, just to analyze it. So for me AI and especially video analytics, there is something important.

And I must say that one of our members that works they have a product RC which is a great illustration of the principle of doing that, they deploy those cameras in schools in order to monitor the vital symptoms of people drink because of pandemic. But when we look to the rest, I mean, 37% of people told us that they use Edge computing for control logic. And that control logic can be in the factory, which is a more controlled environment. But it could be literally in the field for industrial equipment which would be automated or semi-automated.

And there, the main reason why people are leveraging edge computing is that your cultural logic will work even if you lose your connectivity to the cloud. And that's an important concern if you have semi-automated trucks that are driving around the mine site, or something like that. And so what I see going forward is really that the diversity of edge workloads will continue to grow.

And even in the consumer market, for example, we talked a bit about smart homes before, I work in IoT and Edge, and I don't have a smart home, because I don't have the time to maintain my own solution and I don't want my personal data about my daily life to be somewhere on someone's cloud, especially if I have no visibility about what they are using to build that cloud and what they are using to build this little smart home hub that would be deployed in my home. But the interesting thing there is that we see increasingly Edge computing adopted in that market so that my personal data won't leave my own home and not only anonymized reporting will be done to the public cloud and things like that.

So over time, Edge computing will take over more and more sectors of the economy and the IT landscape, generally speaking; because it makes sense to deploy compute and storage closer to the source of the data, and save on bandwidth and minimize latency, and etc. Even in gaming, right? If you're a bit of a gamer, the one thing you obsess about is your ping to the server. And so by deploying edge infrastructure in games, you will be able to minimize that and provide a better gaming experience for everyone involved.

Erik: Maybe just to quickly touch on a topic that wasn't explicitly in the survey, but is maybe somehow referenced here, which is the topic of 5G and local clouds. I've heard people talking about the deploying 5G clouds, dispersing them throughout cities, specifically, or let's say at a stadium or in an industrial park to provide this really high bandwidth, low latency connection approximate to where the data source is. Do you see much of this in action today, or mostly plans and hypotheses about how this might work?

Frederic: The funny thing about this is that most of the things we say 5G will enable, most of them could already be done with LTE connectivity. Theoretically, 5G brings much more bandwidth, but you need to be in very, very specific conditions for that to be delivered. Well, I'm Canadian, so if you think about a huge country like ours, it's not tomorrow that you will go to a mine site in another part of the country and then get 5G connectivity there.

The one thing that we see happening there is really that 5G is an opportunity for telcos to get into the game of application management and delivery. Because right now, when you think about it, most telecommunication players are just providing your pipe that enables your customer to get between their personal devices deployed in the field, and some cloud, and it's the cloud app or skaters that get your most of the value there.

5G is an opportunity for telcos to start deploying application management and orchestration capabilities at the Edge. And then it means in turn that it will be fairly easier a few years from now to be able to leverage those fields deployed little data centers right and left that will be around the communication infrastructures owned by the various telecommunication companies. And this is probably the one game changer that 5G will bring, not so much the wireless innovation when you can make it happen in those very particular circumstances and it requires lots of antennas and everything. But the fact that the business model of at least some of the telco players will change and they will start to refer you to host applications near the cell tower, which in turns mean you will get if you are solution provider you will get much more flexibility about where you will deploy your application components and how.

And in turn, this illustrates the point we were making in our recent whitepaper on Edge computing that we published at the Eclipse Foundation. The team of that whitepaper was called the well EdgeOps, which is our vision of DevOps, but tweaked for the Edge environment. And the point that we were making there is that you need a platform if you are orchestrating workloads like that, that has been built from the ground up for the edge, because the edge environment is so much different than the cloud or data center environment.

To us at the Eclipse Foundation, and especially in our Edge native working group, this is a fantastic opportunity, because you will see infrastructure appear right and left not only in cities, but in more rural parts of our respective countries. And from there, this will enable cheaper and more effective deployment models. Because essentially, the main tenets of EdgeOps is that there's no one single edge, or there are so many potential edges. That in fact, when you look at all the places you can deploy compute and storage from the very, very hedge where your cell tower is, or even where your constrained device is deployed in the field to the cloud, it's a continuum. And so you need an Edge platform that will be flexible enough to enable you to deploy components of your data plane, monitoring plane, and management plane all over the place at any point on the Edge to cloud continuum.

And to us, this is really the way to see this, not thinking in a very fixed ways and saying, oh, there is one edge or two edges or three edges, and there are many taxonomies there, but rather see this as a continuum and focus on platforms that give you the full flexibility to deploy any component at any point on that continuum.

Erik: Maybe you can leave us with a few practical recommendations. So based on the learning from this report, and in your experience more generally, what do you see that companies should be doing today, whether it's specific activities or more directional changes in mindset in order to make better use of IoT and edge solutions?

Frederic: And there, it depends if your organization is a consumer, or adopter of those technologies, or if you're a solution provider, or even a platform or a software vendor. So my recommendation to enterprises, to end users would be to select vendors and service providers that really embrace open source and open standards. Because this shows that your supplier in this case pays attention to the market trends. But not only that, you're not locked in to that specific vendor and you'll be able to maintain things over time, if for whatever reason they drop from the market.

The second recommendation I would make to enterprises is really to start planning, deployments of IoT and edge at scale. The market is mature enough, there is a wealth of solutions available, both commercial and open source, or well, in the case of commercial, I would recommend to at least have a look at the ones that have an open source foundation, obviously, but the ecosystem is robust. So now is the right time to jump in if you weren't so sure about the potential. The potential is there and you have now the breadth and variety of choice to do effective deployments.

Then to solution providers, what I would recommend, usually incorporate open source platforms in everything that you do. Don't waste your time write code that others have already written, and probably in a better fashion than yourself. And obviously, if you are an IoT-focused solution provider, please pay attention to Edge computing, because this is increasingly becoming a deployment model that is common. And there are many, many benefits to yourself and to your potential users. And so the benefits including reduced latency, bandwidth savings, all of that, you want to have them and you want your customers to benefit from them. So if you're a solution provider, it's a no brainer.

Then ultimately, if you are a platform or software vendor, it's important to buy data security and data sovereignty features and solutions across everything that you do. Because people are starting to pay attention for that, especially in not only privacy angle, but if you take into perspective the protection of industrial secrets and all of that it's really important.

And then my second recommendation to platform and software vendors would be to create offerings that optimize specific workflows or mitigate specific challenges. There are so many potential use cases. If you're trying to be too generic, it will be probably hard for you to really sell the specific value you bring. So focus on specific workflows or to mitigate specific challenges and I think you'll be able to build for yourself a road for success. So roughly speaking, those would be my recommendations.

Erik: Yeah, that's a recommendation really resonates with me. And I guess I'm not surprised by now, but always a little bit shocked how broad a lot of companies define their solutions. And I understand it's because in theory, they are horizontal solutions that can be used by a wide range of people. But that's actually not useful when you're trying to bring a solution to market.

There's a company that we had on the podcast a few months ago and what I really liked about their solution was that they focus completely on compressed air, so predictive maintenance for compressed air. And for me, that's just such a great starting point for especially a smaller company, the company's called Ecoplant. But then you have one specific area that you can really master, it's a pain point that many companies have. And this I think, it's a great lesson for especially younger companies that are building IoT solutions, or larger companies that are extending into this market, find a pain point and be the best in the market at it.

Frederic: And we see this in our ecosystem as well. A good example of this would be [inaudible 57:01], which is a startup based in Belgium. And essentially, what they do is that they adopted the main components of the BOSH open source projects. And what they did is that they built a platform, including devices which is really focused on petrochemical plants. So they have devices they install on valves to monitor their physical position and things like that, etc. So they have really a laser focus vision on that segment of the market. And they were able to go to market very fast, because they leverage our open source building blocks.

And then the specific focus may be not for a specific industry, but to a specific type of user. So we have another startup called Cedalo, and what they do is essentially they are bankrolling the development of the mosquito MQTT broker. But on top of that, they build something called the Eclipse Stream Sheets. And Stream Sheets is a spreadsheet like UI on the top of real time IoT data flows. And so non-technical users can use Stream Sheets and type formulas like they would do in Excel or Google Sheets and things like that, but they have access to the real time data flow straight for the factory if they wish so.

And so in the case of Cedalo, their focus is really that type of non-technical user who is used to use a spreadsheet. But really, what we see happening in the Eclipse ecosystem is that you can view every successful startup focused on open source and at the same time with the great commercial success.

Erik: Well, Frederic, I really appreciate you taking the time to walk us through this today. But if some of our listeners are interested in reaching out to you and learning more about Eclipse and your role in IoT and Edge, what's the best way for them to get in touch with either you or the organization more generally?

Frederic: So they can visit our Eclipse IoT website at iot.eclipse.org. And obviously, we have a Twitter account for that which is simply @EclipseIoT. We've got distinct accounts for our edge native and [inaudible 59:15] working groups as well. And if they want to reach out to me directly, I'm on Twitter as blueberrycoder. And obviously, I'm on LinkedIn as well myself. So don't hesitate to send connection requests, I have an open kind of networking policy.

Erik: Awesome. Well, Frederic, again, really appreciate your time.

Frederic: Thank you for the opportunity. It was a great conversation. And thank you so much.

Erik: Thanks for tuning into another edition of the industrial IoT spotlight. If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Finally, if you have an IoT research strategy or training initiative that you would like to discuss, you can email me directly at erik.walenza@IoTone.com. Thank you.