The Co-Inventor of MQTT: Arlen Nipper from Cirrus Link Solutions
Inductive Conversations17 minute episode Apple Podcasts | Spotify | Google Play | PodBean | TuneIn
We’re sitting down with the co-inventor of MQTT to discuss the role of IIoT and MQTT in business, and building secure and efficient architectures. We’re discussing how companies can overcome analysis paralysis, and get people to think differently about connecting decoupled data and devices into their infrastructure. This episode examines how business leaders expecting IIoT to disrupt operations can have a comprehensive strategy moving forward.
"I've had half a dozen integrators come up to me and say, 'Arlen, you've made SCADA fun again. You've made it interesting. You've made it relevant to the millennial engineers.'"
Arlen Nipper has over 38 years of experience in the SCADA industry. Over his career, he has worked for Amoco and Koch Oil and then served as the President and CTO of Arcom Control Systems, and Eurotech Inc. His experience covers a broad range of technology, from the design and manufacture of embedded computer systems to complete SCADA system infrastructure implementations for many companies. Nipper is also the co-inventor of MQTT, a SCADA message transport, in conjunction with IBM, and has over 18 years of experience designing high-performance SCADA systems using MQTT and Message-Oriented Middleware. Nipper was also involved in many of the activities that have recently led to MQTT being a dominant IIoT messaging standard.
Don: Well, let's jump in. First off, MQTT. How'd you get into it in the first place for God's sakes? This isn't your first rodeo, how did you get into that?
Arlen: The background on how I invented, or Andy and I invented MQTT is that, when I came out of Oklahoma State in 1978, I went to work for Koch Oil. Koch Oil had just started putting out all their pipeline systems from the refinery in Oklahoma down to Texas to the plastic plants. I had a lot of experience in putting in SCADA systems, putting in tank form control systems and just SCADA in general. Well after that, I co-founded a company called Nova Tech that were in the business of building protocol converters, because now we're in the late 1980s, early 1990s and we had all of these legacy poll-response protocols and people wanted to upgrade them. They wanted to have newer protocols, and we were basically in the protocol conversion business. Then, if you look at it around the middle 1990s, a very disruptive thing happened to our industry in that AT&T got deregulated.
Now you have to realize that, at that point in time, probably 95, 98% of all of our SCADA infrastructure in the United States, were using multi-drop phone lines as provided by AT&T. They were heavily subsidized, they were motivated to go out and help you troubleshoot your industrial control systems and your communication circuits. Well, after the deregulation, costs skyrocketed, quality went completely out the window. A new technology came in with the advent of the very small dish VSAT systems. Now we had AT&T try them, Goulet, Spacenet, Galot, Scientific Atlanta and everybody was putting out VSAT systems, but now we had a double problem in that everybody had their own proprietary VSAT transport. Now we had legacy poll response protocols with proprietary transports. In that, we invented a protocol with that with AT&T called SNET.
It was the precursor, if you will, to MQTT. Well, in the late 1990s, we were working on a project with Phillips 66, and they had finally gotten a pure TCP/IP-based VSAT system. And we were trying to figure out how to get more information over limited bandwidth and keep the response times even faster from their control equipment out in the field. Having IBM was, we were working on a project around the same time. That's when Andy Stanford Clark and I got introduced, and we basically took the 20 years of experience that I had in SCADA and all of these poll-response protocols and all the experience that Andy had in applying message-oriented middleware technology, so service-oriented architectures. What we know today as enterprise service bus, and we smashed those two things together. And what came out of that effort is what we all know today is MQTT.
Don: That's a great intro story, but I think I need to back up just for a second. Message Queuing Telemetry Transport (MQTT), what is it?
Arlen: Well Don, it's actually quite simple, and I think that's the reason it's still around after 20 years. It's really nothing more than ... there's two components you need for MQTT architecture. You need an MQTT client, and you need an MQTT server. Now the client could be a flow computer, it could be a PLC, it could be an edge-of-network gateway, it can be any sensor out in the field. And it basically can connect into an MQTT server and publish information when it changes. So therein, you get the bandwidth efficiency because you're not constantly polling for things that didn't change. Now, on the application end, applications like a SCADA host system like Ignition, or maybe a data historian like OSI PI can connect in and they can subscribe to the information that we're publishing. That's why these systems a lot of times are called pub-sub systems. Well, publishing process variables to applications that are subscribing to those process variables.
Don: So what we're seeing with that pub-subscribe methodology is that it's very bandwidth-frugal and allows you to do a lot of data and we're seeing things that maybe some people know about, but they don't know why it makes so much sense. But Facebook Messenger, to name one, or Amazon Echo are using MQTT as a transport protocol. They do volumes and volumes of data across large numbers of users, and it almost seems like, even though you guys did this 20 years ago, that there's either a rebirth or an acceleration of adoption or something started happening again in the environment to bring it to the fore. Is that a misperception or is there an acceleration? And if so, why do you think this accelerated adoption is happening now?
Arlen: Well Don, you're right. I mean, for a long time ... at the time, it was Arcom Control Systems and IBM. We had this technology, and we were winning projects, but it wasn't being widely adopted. And then this thing happened here, what, eight-ten years ago called cloud computing, and all of a sudden we had access to these massive amounts of computing information, but we needed to be able to have our devices talk to that.
So at the time, Andy and I had gotten MQTT, we had gotten it into the Eclipse software foundation. From there, we got it through the OASIS Standards Body. So we kind of built the infrastructure. We have standards now, we have open source very mature open source software, but then the Internet of Things happened, and people found out that things like Raspberry Pis, they could start putting MQTT on because it was so easy to implement. It was so easy to deploy.
So now you started seeing this proliferation of things like Raspberry Pis and you could go download MQTT for just about every programming language. From that, it started to take off, and we were starting to get the conversation in the industrial sector. So we started hearing of that, and I think that's why all of a sudden it's like MQTT came out, it disappeared, and now it's resurging over the last four or five years.
Don: Sure. No, I can really see that, but I still ended up, in just my own side reading and stuff, getting a little bit troubled about something that I would like your comments on. While there are gosh knows how many different consortia around this subject of the Industrial Internet of Things or Industry 4.0 or the digital enterprise, and these alliances and these consortia, companies are all coming on board. You go to a conference these days, you don't see anybody's backdrop to their booth that doesn't have the IoT on it in one way or another, and yet I'd like your comment on some numbers that I saw from the World Economic Forum recently that I think talk of a problem that's running pretty deep in the real adoption of real solutions. Their research said that 84% of business leaders expect the IoT to disrupt their operating models in the next five years. That's a pretty heavy comprehensive ... I mean, a very big number say, "It's going to disrupt us."
Then the same study says 7% of those leaders have a comprehensive IoT strategy. They know where they're going. To me, that says, "Oh my God, 73% by this research have none at all." Now here we have research saying there's going to be $100 trillion worth of economic benefit in the next seven years coming to IoT to industrial organizations through IoT. We have all of the possibilities. Everybody knows it's going to disrupt their business operations. In less than one out of ten has a clue of how they're going to accomplish getting that value.
So why the big gap, and how do we approach bridging that gap?
Arlen: Well Don, that is interesting, but the notion here is that what you're saying is that we're going to go from an operational technology base that has taken us the last 45-50 years to build up, and then in a matter of a few years we're going to switch that to a total digital transformation, and we're going to do that in one fell swoop, and I think that's what's scaring most of the customers right now, or not scaring them, but they're in analysis paralysis.
Don: Like how do I start? "What do I really do" is step one.
Arlen: Exactly, and we had talked about this. We were saying, you can't eat the elephant all in one sitting, but you can eat the elephant a small piece at a time.
So as these companies look at technology, one of the things that I like to point out when we're talking about MQTT and digital transformation is one of the ... I did a TED Talk here four or five years ago, and one of the notions I had is that the Internet of People took off because of this thing, this protocol actually, called HTTP. It wasn't perfect, but people could get their hands on it, they could implement it, and that's how web browsers came about, and that was the explosion of the Internet of People.
I think what we're seeing here is the cusp of everybody agreeing that technologies like MQTT ... we're going to bring those into the industrial sector. We're gonna stop talking about all of these poll response protocols. We're going to implement a messaging technology that's going to let us then adopt and address those 73% of industries that don't know what they're going to do next. Because really there's no technology to address it.
Don: Yeah, when you think about that though, you're also dealing with one of, I think, 45 years of embedded thinking in systems and development for traditional poll-response and traditional view of how you architect and build off the topology for some sort of industrial solution.
So I've heard you comment on this before is that we really need to fundamentally get people to think differently about their architecture. They got to stop connecting devices to applications and they got to start connecting those devices, and the data they have, into infrastructure. So that is a shift, or decoupling, of that device from that application is a shift in mindset, and it talks of a whole new approach to architecture.
Can you at least succinctly comment on why that's so important and what it does to facilitate this transformation?
Arlen: Well sure, Don, and again I've been in the industry. We did ... the reason we did poll-response protocols, we did it for the right reason. You know, let's go all the way back to the fact that we had multidrop phone lines that were ... it was like your old party line phone line, right? You had to ring it up, and if somebody else was already on the phone, you couldn't talk. And so we had to invent poll response protocols in the beginning. But in 2018, with technologies like TCP/IP, it just doesn't make sense.
So what we're talking about here is this new architecture is what if you could imagine that you connect devices to infrastructure, and then you connect applications to infrastructure. Best-in-class applications, new applications, and then you've got one too many data architectures. You've got an architecture that's plug and play. You've got an architecture that's efficient, and you've got an architecture that's secure.
Don: No, that totally makes sense. So I have just maybe one final question before we move to wrapping this up, and that is 'Why now?' I mean, come on, you and Andy were working on this 20 years ago. I understand the need has accelerated with the cloud and Big Data and machine learning and all the things people are trying to do, but why now?
Arlen: Well Don, it's funny you should mention that because when I was working with IBM and all of this, we were taking a IT-down-to-OT approach. We were ... IT is coming into OT, and 'Oh you should be doing this, and you'd be doing that.' The problem was they didn't have the paradigm experience, nor should they, of what is a PID control loop, or what is a four to twenty milliamp device, or what are limit switches open on, open on a motor-operated valve.
But in the meantime, we weren't able to show it on a best-in-class SCADA system. And unfortunately, it took Ignition 17 years to basically come out with the fact that they're going to put the best in class in QT technology on their platform. So I think the big thing here is that instead of Arlen drawing pictures on the whiteboard and you trusting me on an 18 month project that this is going to work, we can walk in, ask you what brand of PLC, RTU, flow computer, industrial control machine you've got, and then we can demo it to you in 30 minutes using Ignition.
Don: Sure. Now I have to say this because you've taken us to task on how slow we were to make Ignition before, but you have to realize that when you were 20 years into your career and talking to Andy about this, the key developers that are behind the Ignition platform, with the exception of our CEO Steve Hechtman, those key developers were in middle school and high school, Arlen. So give them a break. They've been moving pretty fast since they got the picture.
Arlen: There you go.
Don: Okay. So all kidding aside, if you take a look at where this is going, and I think probably wrapping this up so that our listeners can get a little closure on our conversation today, but is there any kind of a maybe final thought? I mean, certainly Inductive Automation is very excited about the evolution of the new architecture, the Industrial Internet of Things because Ignition's a platform that plays as a part of the toolset necessary to accomplish that transformation.
But any final words you have for just the audience on the evolution, role of MQTT, where you see industries going? Anything to just wrap up here?
Arlen: Well Don, the most interesting thing to me in the last three years that we've been working with Inductive Automation is this very exciting ecosystem that we've put together. I still remember that at the keynote last year at the Inductive Community Conference that I stated that I've had half a dozen integrators come up to me and said, "Arlen, you've made SCADA fun again. You've made it interesting. You've made it relevant to the millennial engineers." We've got an ecosystem of 35-40 OEMs now that we didn't even know about three years ago. We've got people putting together these systems.
So having this, being able to roll it out, say it's here, it's available, it's IAOT, and we can put it in today. That's what excites me.
Don: Yeah. I have to say that excites me too. It's a really ... it's a fun game to see. So many of these industrial organizations get excited again about what they're doing and to see those integrators get excited again.
Arlen: Exactly. It's not the status quo anymore. It's not the entrenched players driving the market. It's the customers coming to us saying, "I want to do this," and us being able to show them how to do it.
Don: Well listen, I appreciate your time. I know you're busy so thanks for joining us today, and we look forward to more work into the future.
Arlen: As do I. Thank you very much.
Want to stay up-to-date with us?
Sign up for our weekly News Feed.