IoT Podcast Logo

 

Welcome to Episode 26 – The Guest Host Special! Join Paul Bullock, Director- BD, EDJX and Agustin Pelaez- CEO, Ubidots in exploring the fascinating realms of Build Vs. Buy, Edge Computing, Cloud Computing and Celluar-Edge Computing convergence ☁️ πŸ’» πŸ’₯

The episode begins delving into the Build Vs. Buy debate, deliberating the advantages and disadvantages of using either process and discussing the question: When is the right time to use Build over Buy or Buy over Build. We then progress towards a compelling exploration of how Edge computing could have enabled more use cases for the Ubidots Application layer and vice versa. Finalising the episode, we cover a recent and exciting innovation- the convergence of Cellular, IoT and Edge Computing – and how this groundbreaking combination will solve the delay between investment and payback for decision-makers and operators.

Tune in and be the first to discover

  • Paul and Agustin’s backgrounds 00:47​- 04:06​
  • The Build Vs. Buy Debate: When is the right time to use Build over Buy or Buy over Build? 04:06​-09:01​
  • Examples of Build vs Buy projects going wrong and why buy is critical for time to market 09:01​- 13:29​
  • Building Application layers and Edge Computing layers: Exploring instances where Edge Computing may have enabled even more use cases on the Ubidots platform and vice versa 13:29​- 18:57​
  • How Cellular and Edge Computing convergence will revolutionise content delivery and mobile user experience 18:57​- 24:05​
  • How Edge compute, Cellular and IoT enablement will solve the delay between investment and payback for decision-makers and operators 24:05​- 27:39​
  • New projects to look out for in Cloud and Edge computing 27:39​- 30:06​

More about the Guest Hosts:

Paul Bullock is the Director of Business Development at EDJX. As an expert within the field, he has a background in M2M, IoT and all levels of the network stack. In particular, he is specialised in product and business development, helping start-up Jersey Telekom and was previously director of BD at Arm.

Agustin Pelaez is the CEO and Co-Founder at Ubidots a business operating a data-driven IoT Application Platform to innovate IoT Application development. As a highly experienced Engineer, Agustin started engineering a monitoring solution for Airbus Germany, but with his co-founder –Gustavo – fathered the Ubidots platform to reduce waste and empower economisation of production resources.

Episode Transcript

Paul Bullock:
Hey everyone, thanks for coming today. Welcome to The IoT Podcast Guest Host Takeover. I’m your host, Paul Bullock. My background is in M2M and IoT, and all levels of the network stack in product and business development. Most recently I helped start a business at Jersey Telecom, their IoT connectivity business, and I was director of business development in the IoT group at Arm, and I work for a really cool company right now called EDJX, E-D-J-X.io, where we’ve developed our own Edge OS and edge network for advancing the cause of edge compute in IoT and other places.

I’m joined today with my guest, Agustin Pelaez. Agustin is the CEO and co-founder of Ubidots. It’s a business operating a data-driven IoT application platform to innovate IoT application development. Agustin started engineering a monitoring solution for Airbus Germany, but with his co-founder Gustavo, fathered the Ubidots platform to reduce waste and empower economization of production resources. So Agustin, welcome. Thank you very much for making time today.

Agustin Palaez:
Thank you, Paul.

Paul Bullock:
Tell us a bit about your background in IoT and how and why you started the business.

Agustin Palaez:
Of course. Well, very glad to be here and very looking forward to learn from your experience, which I think is very rich, both in terms of business and IoT. So looking forward to have a nice chat. Well, I got into IoT because of my co-founders who … Well we’ve been together, if I sum up all the projects we’ve done, is going to account for about 10 years. At the beginning, we began doing projects for the healthcare industry. So, I’m originally from Medellin, Colombia and there were a lot of hospitals in our region that had legacy systems to monitor temperature and humidity in the vaccines refrigerators. So we began creating IoT solutions for them and in the process we thought, “Well, it would be very cool if there was a place where we could just send all of these data without having to deal with software engineering teams and requirements, specifications, visualizations.”

Back then, we couldn’t find anything, we’re talking maybe like seven to eight years back. So that’s how the idea of Ubidots was born and how you like to call it in entrepreneurship, the motivation to scratch your own itch, solve a problem that you had first, and that will give you a lot of tools to understand what’s needed out there. That’s exactly what we did. We built Ubidots as a easy means for any IoT entrepreneurs like ourselves back in the days to create entire IoT projects without having to hire a software engineering team.

Build Vs. Buy: The Debate

Paul Bullock:
Cool, the industry really needs that. IoT is not a very mature industry, it gets lots and lots of press everywhere, but there’s all kinds of little companies like yours and the one I work with trying to put things together and buyers, whether they’re consumers, or enterprise like Bosch, they always face this challenge of build versus buy. I think it’s fair to say that most manufacturing processes and procurement people in IT, people see the potential of IoT, but they keep figuring out how best to do it. Do you have a perspective on build versus buy in the context of Ubidots? Maybe you could point to an example where you’ve built something that gets re-used by more than one customer and thereby creates inherent efficiencies?

Agustin Palaez:
Absolutely. And I think the build versus buy discussion is very common among us engineers. A lot of the buyers on the other side are trying to build things themselves because, well, we tend to do that as engineers. We like to download the latest Grafana version and install it in our own servers and use that to create a quick prototype, to visualize sensor data. I personally do that myself and I love it.

Paul Bullock:
Cool, yeah.

Agustin Palaez:
It’s fun. But when it comes to real on-production application, there’s a lot of hidden costs in the process. So one, of course, is the time that it takes to build the entire stack, even if you use existing tools. The second is when things go wrong. So you can have like this mashup of, for example, Grafana with no red, with maybe a layer of storage and some alerts using Twilio, and it might work well for a year, but then sometime at 3:00 AM in the morning, it’s going to break for whatever reason and then you’re not going to be there to fix it and this hour or extra two to three hours that it failed might represent a huge loss per production application.

So I think it’s great to think of with a lot of the tools out there and from our perspective, what it means is Ubidots put together a lot of backend and front end ingredients that are common to both sides of the applications. So, from time-series storage, to alerts, to connecting with other systems through web hooks and APIs, of course, visualizations, create your own widget, or use one of our 20 existing widgets, create several less functions to compute very rapidly a specific transformation about the data, et cetera.

So, we are a tool at the end, we’re also one of those tools, but we packed the most common ingredients at the application layer. So yeah, in the build versus buy description, we would definitely be the buy one where you can just buy one license per month that’s going to solve you a lot of headaches.

Paul Bullock:
It’s funny, everything you said there echo is a lot of what we do at EDJX, but from a slightly different angle. Like we see, and we saw this at Arm a lot, I talk to so many cool, smart guys, but everybody’s trying to reinvent the same wheel over and over again.

So at EDJX, what we’ve done to tackle this is develop our own OS, but arguably more importantly, our own edge network protocols, which are comprised of distributed ledger content addressing and fundamentally a peer-to-peer network, which means any compute power, doesn’t have to be ours, that joins our network automatically becomes part of the edge cloud, so to speak. When you published an application like yours onto the EdgeNet, it becomes simultaneously available at every node.

Agustin Palaez:
Awesome.

Have you seen any examples of Build Vs. Buy projects going wrong?

Paul Bullock:
So there’s this inherent multi-tenancy to it, which means, in the same way that AWS is multi-tenant, but you got to pay Amazon arms and legs to use their stuff, we’re offering a multi-tenant platform that costs of finger, as opposed to an arm and a leg, and most importantly, it brings the compute power as close to the data, whether that’s a thing or a person, as it wants to be. Our stuff really does echo each other.

I mean, I had a related follow-up question for you. If you can think of an example where you’ve seen customers, or you’ve come into a project after something’s gone wrong, or maybe you’ve been on a project, lost it, and then saw it go wrong afterwards, can you think of an example like that?

Agustin Palaez:
Yeah. You mean within the build versus buy example?

Paul Bullock:
In that frame, yeah.

Agustin Palaez:
Yeah, so we’ve seen instances where they come to us after they failed. So 20 to 30 people company, they already tried to build something on their own. They had some developers in-house, other developers hired to a freelancing platform, for example, and after a few years, well maybe after a year or maximum two years, they said, “Okay, we’ve used all of these resources, but more than the resources, we are not in the market yet.

So we wanted to have this [inaudible 00:09:47] tech solution because we had this many opportunities and now we’ve lost a few of them because of the time to market.” So, now they come to us and they realize they can build maybe 80% of the things real quick. There might be a remaining 20% that’s a sacrifice because you are using a product instead of building your own solution but this 80% to 90% already takes them to market.

I think that’s very important. I’ve seen it again from a failed project too going into the product and I’ve also seen it the other way. There was a hospital in the Philippines that had been a customer for a lot of years. I realized they began offloading a lot of devices, like okay, something’s going on here. Maybe they don’t like your product, maybe they’re shutting off the project.

But we reached out and it turns out they decided to build a platform on their own. They had very specific requirement, including deployment within the Philippines for health regulation issues. They said, “Well, we know we can get away with $200 per month that we pay you but we know this is going to cost us a few thousand and a few months or years, but we want to do it because it’s a specific regulation.” So, I’ve seen both cases. I think both are equally valid.

Paul Bullock:
I think a couple examples in industrial IoT where people are trying to do industry 4.0 and all that and invariably it means you have to drop boxes into say, manufacturing floors. Those boxes are probably IoT gateways of some type and they run Windows or something. So these projects in variably run into the challenge of trying to coordinate so many different pieces of a given enterprise, whether it’s their product people or software people, their hardware people, their IT people, et cetera.

That really loads a lot of cost and pain and time into the deployment and ends up being a frustrating thing, when what we’re doing is .. Again, I’ll use the AWS example, as opposed to hiring, you want to accomplish something, you can hire some devs, write an app on AWS and deliver a business requirement. You don’t really have to involve your IT department so much.

We’re replicating that to a pretty meaningful degree in that getting some very vanilla compute in place, installing Edge OS, writing your app yourself on Edge OS, or getting somebody to do it and then distribute the vanilla compute wherever it needs to be and the app just appears there and you can have your back haul via cellular or plug it in with ethernet, whatever you want. It’s a way to reduce a lot of the friction that you get in trying to change enterprise direction in deploying new things.

Agustin Palaez:
No, I see it and maybe if we had known about you guys, when this Philippine hospital [inaudible 00:12:58] we would have been able to retain them by using your distributed cloud.

Paul Bullock:
The very next one you come across, all right?

Agustin Palaez:
Yeah, sounds good.

Are there any instances where Edge Computing could have enabled more projects than Cloud?

Paul Bullock:
So Ubidots is a pure cloud software play, actually what you just said is half answering the question I’m about to ask. I’ve actually written down here, “There might’ve been an instance where edge computing could have been [inaudible 00:13:25] even more projects.” Except in the Philippines [crosstalk 00:13:29].

Agustin Palaez:
Okay, yeah. I was getting into that, I see. Yeah, there’s a lot of instances. I can think of a few. I think, well, what you just said about industrial IoT is one, I think there’s the ecosystem, like control automation, this type of system integration and the end user is very used to buying things upfront, maybe not very friends with recurrency yet, and along the same lines, they like to host a lot of things themselves. So a great pitch could have been maybe, yes, I mean, we have the Ubidots cloud as a way to see the last mile of the data, but you can keep it closer at the edge for your own analytics, for example.

Specific to that could be, they measure a lot the productivity, the OEE for instance, which is a combination between equipment downtime, availability, productivity, and quality. So they like to measure, every factory in the world does it, they just measure it either very manually or more digitally.

So, we see a lot of use cases where they measure productivity and availability, no questions asked. It can go to the cloud, whether a machine is turned on or off, how many units you produced, but when you need to bake in the quality aspect, that’s more tricky. Sometimes you need the image processing, imagine a production line, that’s checking whether each unit looks exactly the same as a reference unit, at a speed that could just not be sent up to the cloud to do an image processing and then come back and say, “Hey, yeah, this unit, it’s okay. Maybe those five to 10 seconds, or even less, it won’t just work for the production.”

So I think that that image processing use case could be a very cool one where you can install some processing capacity at the edge that responds faster for this, whether a unit has enough quality or not. Then latency in general, our main infrastructure is in Toronto, IBM Toronto, your home city, I guess.

But when a customer is in Australia, for example, then their latency becomes more apparent, maybe it’s 500 milliseconds more than it should. So in those cases, having a satellite ingestion tier where you can reply, like the API can respond faster to devices, maybe 300 milliseconds as opposed to a second, or even less, it would’ve made sense in some use cases in Australia.

Paul Bullock:
Yeah. As you’re saying, I think that that context generally might be video as a sensor, where you want to capture images and make decisions in real time without bearing the cost of back-calling it to some computer somewhere, a zillion miles away.

Agustin Palaez:
Yeah. Now another one came to mind now that you said predictive analytics. So, very tied to vibration in a lot of machines that we work with, ncd.io, a very cool gateway and hardware manufacturer in the industrial space. So they already, and a lot of other players, they already offer like vibration sensors. The problem is that it’s a very high frequency. So most of our use cases are on a permanent basis, update the variable.

Paul Bullock:
That’s a great use case for edge because the sampling frequency is negated by latency.

Agustin Palaez:
Exactly, yeah and if you are doing a fast four year transformation to know whether a machine is vibrating at 80 Hertz, as opposed to 75, which it should, maybe it’s because it has a problem, it needs maintenance but right now we don’t have those type of use cases that actually send the entire acceleration time series, because it’s a very high sampling rate.

So if there was something on the middle where you say, “Okay, this millisecond spaced acceleration data is 75 Hertz.” Then you send only that data to the cloud as opposed to sending a thousand samples per second, so that’s a very cool one as well.

What trends are you most excited for this year?

Paul Bullock:
That’s a good one. We should think about that offline actually. I wanted to hear your thoughts, and I’ll share mine, on what’s the coolest thing happening this year for you in our industries?

Agustin Palaez:
I am very excited about cellular IoT.

Paul Bullock:
Me too.

Agustin Palaez:
But I think you’re more of an expert in that regard.

Paul Bullock:
No.

Agustin Palaez:
So it will be actually more interesting to listen to your inputs as well. I have some initial thoughts but, I don’t know, would you agree that that’s a big thing these years?

Paul Bullock:
I think it’s pretty huge. I think there’s an intersection happening between edge compute and cellular in a couple of different ways. Around the things that we’ve been talking about, getting computer closer to the people or things, compute always wants to be as close as they can get to the raw data, but there’s a thing called neutral host happening and neutral host is a standard that allows anybody to put … You’ve probably heard of a thing called private LTE, right?

Agustin Palaez:
[crosstalk 00:19:36].

Paul Bullock:
I can put up my own private LTE network in my house if I want, and I’m on my own network. They also have a concept called neutral host that takes a little subcomponent of a big mobile network called an eNodeB and a little radio and I can send up a signal that’s a 4G, 5G signal and users can actually … an AT&T user can roam onto my little network. That’s cool at one basic level, where the operators can extend their signal, very inexpensively into places of congestion, new builds, whatever it is. But imagine if you throw a bit of compute on there.

So I’ve got this node that’s maybe a 22 use rack, a half rack server with an eNodeB and a radio mask on top of it that’s putting out a 4G signal, all the users and things can log onto that. I can host applications locally where the users of those applications are mobile-centric, but the killer is actually … One of the killers here is content delivery because 50% of the world’s internet is delivered over things like Akamai and Fastly and Cloudflare, but content delivery systems for mobile don’t exist. The way data is moved around mobile networks is a nightmare.

There’s a massive trombone between your phone and the sensor of the mobile network and moving content delivery closer to the mobile data consumer is a net new application of both neutral hosts like I described, where I’m pushing the network edge out, the radio network edge, and I’m also pushing the compute edge out. So you can imagine in say, like a Walmart, or a Asda, or a big company building, when I can extend AT&T’s network in there and bring a bunch of application services and bring a bunch of content all in one 22 use server-

Agustin Palaez:
Wow.

Paul Bullock:
… I’m solving a lot of problems and creating several new revenue streams around that edge node. So, I think there’s something in there.

Agustin Palaez:
That’s very cool. From the IoT perspective, I always thought about the typical energy consumption range, price per month, but I never thought about what happens in the backstage or at the edge of a cellular network. What you just mentioned just opens up a lot of faster applications and even more, these startup opportunities I guess. It’s going to be exciting to see what happens in next years.

Paul Bullock:
Yeah, we’re not the only ones thinking about this, but once you get your head around being able to put your own spectrum canopy up, charge the MNOs, the mobile operators for access to it to improve their levels of service to their customers, put applications and content into the edge node that allows the location of that node, whether it’s retail, or corporate, industrial to make use of edge compute. I think I counted five different revenue streams that can be built around one edge node. Plus, it’s making stuff work better in general.

Agustin Palaez:
Yeah and out of those five revenue streams you just mentioned, maybe one or two are directly IoT-related, so there’s even more. I always approach this cellular IoT from the IoT perspective, but there’s even more to it I see, to end users, consumers. It’s very cool.

Paul Bullock:
Yeah, it’s a bit grand but it’s a bit of … We call it the edge economy, which I think is a bit grand, but there is an edge economy to what’s going to happen, in the same way there has been a web economy.

Agustin Palaez:
Very cool and one more reason to track the progress of cellular IoT. What fascinates me is, of course, the tech behind and everything and the enablement for a lot of IoT applications that we potentially see within our user base. A lot of them would like to take the tracking to the next level. They already track cars and equipment, but maybe they want to track cheaper equipment or smaller equipment. I think it’s fascinating because from the economic point of view, what’s going to happen? What are those decision makers going to finally decide?

I’ve seen in one side of the spectrum, you see a big push in cellular IoT, which makes sense. But in the other side, you also see NTT [DoCoMos 00:24:36] from Japan, they actually shut down the NB-IoT network last year, which was strange to me. So I’m thinking. “Okay, what’s going to happen?” What I see in Colombia, and maybe in other countries in Latin America, is the operators are a bit shy to make the final decision to deploy because what they say is, “Okay, we have all of these customers that already pay us for,” I don’t know, ” … fleet tracking and all of these applications, they’re paying us maybe $3 per month,” to say any value. “And now you’re telling me we need to upgrade our network, do these millions dollars investment to be able to charge less per device. So what’s my incentive?” Of course the incentive is the scale, but there’s a delay between the investment and the decision and the actual payback, which if we learn from SIGFOX and other players, this payback might be longer than we all expected. What do you think?

Paul Bullock:
Right, I agree with you, right and I think it’s a chicken and egg argument. The CFO at your typical mobile operator does not want to be the chicken waiting for the hatch.

Agustin Palaez:
Okay, that’s a good way to say it.

Paul Bullock:
So you have a disconnect there and that’s another dynamic that excites me about that conversion, so cellular IoT and edge, is that this neutral host model can solve this, to not every degree, but to a substantial degree for these operators. The operators hate 5G, because it’s an opportunity for them to double their radio access network costs with no clear chance of any new money left.

Agustin Palaez:
So it’s your job to show them new revenue streams, I guess?

Paul Bullock:
Yeah, yeah and ways to reduce their CapEx by attaching revenue to smaller radio nodes that they can participate in too, but then other participants within the range of that signal, there’s many other participants, either providers of services and content, or consumers thereof.

Agustin Palaez:
Very cool. Well, I look forward to tracking … We should do this again, like in one year, to see.

Paul Bullock:
Yeah.

Agustin Palaez:
To see their nodes.

Paul Bullock:
Well EDJX has a couple of cool PR announcements coming up, so do follow our newsletter if you can.

Agustin Palaez:
I will.

Paul Bullock:
I don’t know, I think … I didn’t have a lot of other items to discuss on this particular episode of the podcast. You guys got any particularly cool projects you want to talk about or do you want to make the audience aware of things they should be talking to [inaudible 00:27:29] about?

Agustin Palaez:
I think a lot of the learnings in IoT is what you just mentioned about … Well, we talked about not building everything from scratch, and that relates a lot to integrating. At the end, if the buyer is finally being aware that they don’t want to build everything themselves, at least they want to see a degree of compatibility between all the layers of the stack. So, we are working on that very actively.

We just launched a new bottle called Ubidots Plugins, which exactly pursues the best simplicity in the industry in terms of connect weather, data sources, satellite data, Laurel One Network servers, potentially tomorrow, if cellular IoT ends up having a middleware that we can also connect too, then we will do it.
So it’s called Ubidots Plugins and we have AWS IoT as a data source. We have OpenWeather, which allows you to drop a pin anywhere in the world and get the data. We’re working on an air quality one, because this time of the year, particularly in my city, there’s a lot of fog and people are very concerned about that, so we want to deploy it soon. Helium, The Things Industries. Yeah, we’re working on a lot of plugins and I think that’s going to be a major leap for a lot of our customers, to be able to enrich their IoT application with data sources that they didn’t know they could tap into.

Paul Bullock:
That sounds cool, man. I like all the environmental monitoring stuff. Well, I guess I want to say thank you to the audience and thank you to the people at The IoT Podcast who allowed this takeover to happen. Thank you to Agustin from Ubidots and thank you to my employer, EDJX for allowing me to do this.

Agustin Palaez:
Thank you.

 

 

The IoT Podcast Team

The IoT Podcast is powered by Paratus People, a leading organisation in IoT Talent Solutions.

Innovation is at the heart of IoT, it is our passion to explore and learn more about this fast paced and transforming sector.

Connect & Get Involved

Your subscription could not be saved. Please try again.
Your subscription has been successful.
Subscribe to our newsletter to be amongst the first to find out exclusive information about The IoT Podcast.

We use Sendinblue as our marketing platform. By Clicking below to submit this form, you acknowledge that the information you provided will be transferred to Sendinblue for processing in accordance with their href="https://www.sendinblue.com/legal/termsofuse/">terms of use