The Co-Inventor of MQTT: Andy Stanford-Clark from IBM

We’re discussing how MQTT got started and how it has evolved over the years. Hear about the big technological developments impacting the industrial sector and what it takes to accelerate the adoption of real IIoT solutions. We’re coming together with the co-inventor of MQTT to combat common problems and major changes within the industry.

“People are sharing templates in the cloud. People are sharing algorithms that they know solve particular problems. Just making that knowledge available so people don’t have to reinvent the wheel over and over again, means all these businesses become more agile, more lean, able to deliver better service to their customers at a lower cost... That collaboration and openness, over the last 15, 20 years has really been more and more apparent.”

Guest Bio:

Andy Stanford-Clark is the Chief Technology Officer for IBM in UK and Ireland. He is an IBM Distinguished Engineer and Master Inventor with more than 40 patents. Andy is based at IBM's Hursley Park laboratories in the UK, and has a long background in Internet of Things technologies. He has a BSc in Computing and Mathematics, and a PhD in Computer Science. He is a Visiting Professor at the University of Newcastle, an Honorary Professor at the University of East Anglia, an Adjunct Professor at the University of Southampton, and a Fellow of the British Computer Society.

Episode Transcript:

Don: Andy, I really appreciate you taking time to sit down and chat a little bit. I just got a chance to meet you this week. You were here at the Ignition community conference. You were also invited in by Arlen Nipper, who I know you have a history with and have worked with. I think maybe you're very well known across the world, and everybody already knows you, but in case somebody doesn't, a little bit of background on what you do. And then maybe bridge that into what you and Arlen got up to, and how you guys work together.

Andy: Okay, sure. Hi Don. It's great to be here, and fantastic week here at the ICC conference. So my background is that I worked 20 years in what we now call Internet of Things. It's been called lots of things along the way. SCADA, depending on which side of the pond you come from. Pervasive computing. Ubiquitous computing. Machine to machine. IBM tried calling it Smarter Planet a few years ago, but the world settled on Internet of Things. Now, of course, we have the Industrial Internet of Things as well. So I've been working that area for 20 years. Back in 1998, Arlen Nipper and I co-designed MQTT together, and I can talk more about the history of that 20 years, later on. But really the last 20 years I've been working IBM's Internet of Things organization, working with large corporations to get their pilot IIoT projects up the ramp of scalability. Trying to get away from a couple of Raspberry Pis, up to the thousands in a proof of concept, up to millions in production. Washing machines, elevators, cars, roll off production lines, and actually go out there and start sending back data.

Now, coming up to date, after the last year, I've been the CTO of IBM UK and Ireland. I work across all of IBM's technologies. Not just Internet of Things. And have a fantastic time talking to our clients about Blockchain, and augmented reality, and virtual reality, and quantum computing, and mainframes and IIoT and how it all comes together in the sophisticated solutions that we're now finding in the world around us.

Don: That's fantastic. It's very exciting. I know this is probably beyond the technical skills of most of the people who are listening here, but I have a question because I know we come from different sides of the pond, as you say, even though my wife is British. But I will say, I need to ask you, what is a bacon sarnie? What is a bacon sarnie?

Andy: Ah. You've been reading my Twitter feed obviously. 

Don: Yes. Of course.

Andy: So sarnie is just a sort of slang name for sandwich. So bacon sandwich is two pieces of bread with bacon in between, which if you have the chance to have to a cooked breakfast, and that we'd quite often eat for breakfast. Not every day, of course, but out when I'm traveling, if I get to a station and half an hour to kill before a train, then the bacon sarnie on the platform is the perfect way to start the day.

Don: And you've done that more than once, I'm sure. 

Andy: I've certainly done that more than once.

Don: Oh my gosh. That's fantastic. Okay, good. We got that technical question out of the way.

Andy: Glad we sorted that out.

Don: Finally. Just a little bit about ... I'm going to talk a little bit about the Ignition Community Conference, and as your first time here, and your experience, but maybe you could talk a little bit about MQTT. Enough to just define it, and give our audience a little sense of what that is and how it relates to the Industrial Internet of Things.

Andy: Sure. Well MQTT is a messaging protocol for the Internet of Things. It's a way of devices talking to other devices, and to applications on the Internet. It's designed to be very lightweight, in the sense that it's very easy to implement on small devices, and it's very light on the networks. So it's non-chatty, and it uses very small messages as far as possible to do its thing. You can think of it as a bit like a postal service. You package up your parcel, wrap it in brown paper, stick, and address on it, and sent it into the postal service, and they deliver it from A to B for you. You don't need to worry about exactly how it gets there. You just worry about what will you send, and what will you do with it when it gets there. That's what MQTT does for you. It will take your data. It doesn't care what you send. Payload agnostic, as we say, and it sends it off from A to B, and then B knows what to do with it when it gets there.

Don: Good. That's a good summary for people. I appreciate that. So you mentioned that you were CTO of IBM UK and Ireland. I'm a little interested in maybe, this is a very critical, been around a long time company, doing a lot of progressive things. What are some of the big technology developments? You mentioned a few things you got to work on very quickly there. That you're working on. That you're keeping your eye on. And that you are maybe most interested in, that have the greatest potential for impacting the industrial sector.

Andy: Well, so we're seeing some really big transit moments. The main one is the industry 4.0 thing, where rather than being built on automation, business is now being built on the data that automation provides. So data from sensors providing what we call the three Ps, performance, productivity, and predictability. So using data to be able to make the business more repetitive. Understanding why the things that go well, go well. Why the things that go badly, went badly. And doing things like predictive maintenance so that hopefully plants won't break down, they'll get an engineer to them before they break down, because the system alerted them to it. That's all about using statistics, modeling and machine learning and increasing the AI to get better insight into the data.

I have presentations called It's All About the Data, because that really is what it's all about. That's what we've seen with the conference this week, is that Ignition, and Sparkplug and the ways you can send data to the cloud and doing machine learning, insight ignition, and all that kind of stuff is all about making use of the data to get those actionable insights, so in other words, you should do this now, because that's what the data is telling us based on what's happening over there out with the sensors.

Don: That's great. Now you mentioned the conference this week. And since this will air in a little bit, everyone won't know what conference you're talking about, so I'm going to tell them you're talking about the Ignition Community Conference. I'm also going to say, because you mentioned it to me, is that you didn't have any understanding of Ignition. And also, for people who don't know, because we cover a variety of topics on this podcast, that don't necessarily have to do with just what we're doing in inductive automation. But we do have the Ignition industrial application platform. 

You have great, deep knowledge of what you're doing at IBM, of Watson and of the industrial sector, and all these technologies, but you had never heard about Ignition. So you walked into this conference Monday morning, going to do a presentation on MQTT 20 year anniversary, not knowing anything about Ignition. But you told me a little story that I'd like you to share with the listeners here. What was that experience like, and how did you get your introduction to Ignition? What happened?

Andy: So I'd been hearing from Arlen Nipper about how MQTT was now using Sparkplug, and Ignition and Ejectors and these kind of car oriented puns. I didn't know quite what he was getting at. I knew it was something to do with industrial automation, obviously, and I knew it supported MQTT, which I obviously know a lot about. So I walked into the conference as you say, Monday morning, to the workshops. Still a little jet lagged, it's fair to say, after my long flight from the UK. Went into the service link workshop, which is all about MQTT and Ignition. So I thought that would be a good way to ... given I already know about MQTT. Good chance to find out about Ignition. The prereq was that you have Ignition installed in your laptop. So that probably counts me out of doing anything hands on this session, because most things take an age to install.

But the instructor showed me where the link was to download the free trial. Six minutes later I had Ignition installed and running on my Mac. No problem at all. Didn't know what to do with it yet, but at least it was there and running. The designer was there, and it was all waiting to make it to go, so that was pretty cool. Through that session I was able to understand how a device can publish MQTT messages using Sparkplug, to specify the topic space, and the payload format, adding all the additional metadata that makes SCADA system come to life. And I had my device publishing in and it was seeing pump values and both little GUI, and I thought it was pretty cool. I kind of started to understand the relationship between the SCADA, HMI, and the MQTT side.

But then in the next session after that, I turned to listening in the conference, and I was sitting next to Arlen Nipper, and I had my laptop on my lap, and was playing around with the Ignition stuff because it was there. And very quickly had the Node-Red Sparkplug node downloaded and installed.

So I built myself a little virtual temperature publisher so it was a temperature sensor in Node-Red publishing data point of a random temperature from 0 to 100 every five seconds, through Sparkplug, going into my local copy of Ignition. I could see the tag come up in the tag list. So I built myself a little user interface. I had a little thermometer thing going up and down. So it was dancing around. And it was pretty cool. So I did and end-to-end SCADA integration, as far as I was concerned, and I did this whilst listening to a conference presentation.

Then the afternoon, it was Kevin and Kathy did the session on machine learning inside Ignition. That for me brought together two things I was not really very well experienced with. It was kind of perfect storm of really taking me out of my comfort zone. There's Ignition, which I knew nothing about. Python which isn't my strongest language, and machine learning which I kind of know a bit about, but not enough to use it in anger. By the end of that session, and having been taken somewhat out of my comfort zone, I did have the examples thing working and I was doing Kmeans classifying of data points coming in from a data system. That felt really good. I could really see the power of the Ignition platform to get insights from data using powerful machine learning algorithms.

Don: That's really a fantastic story. I really love when you told me that. I'm glad you shared it, because we say here Ignition is user friendly. Ignition is intuitive, and you can jump in use ... and you just jumped in knowing nothing Monday morning, and you were doing stuff that day, and the days beyond.

Andy: One thing I thought was pretty cool, which I've never seen on editor before was that when you're creating user interface, the live data actually comes in there and then. So you actually see the thermometer going up and down while you're still dragging it around the screen. I actually didn't realize there was a run mode. I just thought that was it. Once you finish dragging things around, that becomes your UI so I was pretty surprised because I thought you needed to press the green button, the green arrow now to make it run. Okay.

Don: And there it runs. 

Andy: Yup.

Don: That's great. So I want to go a little bit different direction. Back to your MQTT experience a little bit, can you give a little bit of sense of how it started and how it's grown over the years, and what impact it's having now? I know there are products like Facebook messenger that are using it. I don't know that our listeners necessarily know the impact that MQTT has, on empowering IOT, and the industrial internet of things.

Andy: So going back 20 years, MQTT was originally designed to be the first open, royalty-free, public domaine SCADA protocol from gas pipelines, and Arlen and I designed it really working with Phillips 66 to solve a particular problem using publish and subscribe messaging on their pipelines. We didn't really think it had much of a life beyond the SCADA world.

So we tried to promote it. We had the Arlen and Andy show, where we'd go around the world trying to create a marketplace for this new technology we'd invented. Mainly the oil and gas companies here in the Midwest U.S. We got some successes, but it really wasn't having the snowball effect we'd always hoped for. I used to joke that our modest plan for world domination was that one day first of all, all devices would talk TCP/IP, and those devices would be talking MQTT to communicate with each other. Now back when we made that prediction in 98, that all devices would talk TCP/IP was really quite a bold step. Now it's laughable, because of course everything is TCP/IP. But back then some of our customers didn't even know how to spell TCP/IP.

I remember giving half hour lectures to some of the customers we met, about TCPI 101. And I thought, I just can't believe I've having to do this sort of not high school, but sort of university level lectures to people to explain how the lower level ... I haven't even started talking about MQTT, or the business applications that it can solve yet. We were right down in the weeds with this stuff. That itself was a bold step because of course poll response across lease lines was pretty much the norm for automation systems. That package switch circuits were starting to come in. VSAT networks were TCP/IP and that kind of stuff. That was a very different world back then.

Roll forward 10 years, and I gave a talk at a conference in the UK where I talked about the cool things I was doing with MQTT, so home automation, and tracking, and doing all these kind of cool stuff. And underpinning it was MQTT. Someone in the audience, because it was an open source conference, asked would the IBM lawyers be upset if somebody were to create an open source version of the MQTT broker. I said, "No. Absolutely not. Between you, me, and the 150 other people in this room, we'd absolutely love that because that would prove that it wasn't an IBM proprietary thing." Even though it was all public domain and open source, people perceived it as being an IBM thing. Having an open source broker would break that mystique.

So there's a guy called Roger Light in the audience. He was really inspired by what he'd heard at my keynote, and he rushed off home and registered the name Mosquitto, with two t's. So it's got MQTT embedded and the name. Absolute genius. He wrote the first open source MQTT broker. And that really was the tipping point. As you mentioned, Facebook messenger, over 800 million clients using it. It later became an OASIS standard, and later an ISO standard, so it's ISO IEC11, 20922. So that got us into a whole load of areas where people said, "Oh yes, but it's not a standard." But trouble is, it is a standard. So we can get into a whole lot of government and utilities where they require a standard sort of gating factor. Now we have literally hundreds of millions of devices.

All the big IOT platforms from IBM, Amazon, Microsoft, and all of the commercial broker implementation get all support MQTT as their primary protocol. I'm very proud to say, our plan for world domination is actually come true, and it overtook last year, MQTT overtook HTTP as the most used protocol for IOT devices on the Internet. So basically, we've won.

Don: You won. And since you had the presence 18 years ago to predict your world domination would come to fruition, and it has, I'm going to ask you to predict a little bit here. So you see how MQTT fits into the world now, and the world of IOT, and IOT industrial internet of things. So, how do you see it, if you were to day look forward over the next five years? How does it fit into the world that is emerging now over the next five years from your view of the world?

Andy: That's a very interesting question. I now have a different viewpoint because I'm not something much only IOT focused. I see IOT in the context of all the other things customers are doing to try and solve business problems. And certainly in the pursuit of it, it being all about the data, and those actionable insights. So the ability to use analytics and cognitive technologies to get insights from data is, that's just going to increase, and increase. People are really only at the start of that journey. As people move more of that processing into the cloud, and they use more and more off the shelf analytical functions to enable them to get insights into data, we are going to see much more data driven businesses.

Another area that we're seeing a lot of interest in this Blockchain. Some people laugh, it has to do with Bitcoin. But Blockchain is really showing up as a key technology, particularly in IOT where any form of goods have been transferred from one person to another, so custody transfer type applications. Where people in the supply chain don't necessarily not trust each other, but they don't need to have visibility all the way down the supply chain for the whole thing to work end to end. So you have local trust. But if you start to look at track and trace, where you've got to trace back up the supply chain to, for example, find out which cow on which farm a particular batch of beef came from because some people got sick when they ate the steaks, you got to work out better trace back to how that ledger, that log of everything that happened set in stone. That's a really valuable thing to have. If it doesn't take much to put the data onto that Blockchain, that becomes very valuable. This is very exciting.

The other really exciting thing we're seeing is augmented reality and virtual reality. VR's crossed an interesting tipping point in that the equipment is both affordable, and it doesn't make you feel ill when you use it. So that was a really key turning point, because you get motion sickness when you use these things. But now it's stabilized, so some of the things, oh you could do that using VR. Sort of engineering applications like taking an engineer to a sequence of operations, particularly as the older generation starts to retire, and the people who built the plants are no longer around to instruct the young people how to fix that particular pump or keep up secure sequence of operations working optimally. They can train a system which educates next generation using the cool, trendy VR headset. So we're seeing a lot of that.

Don: That's great. Now you also have another title. I know you're chief technology officer for the UK and Ireland for IBM. But I think you also have the title of master inverter that goes with your name too. And IBM is a company that's having a significant impact across industries overall, but how do you see IBM and the work you're doing there, and the newer technology you're developing? Lots of people know about IBM, but they may not know much at all about that. How it plays out in the industrial sector. How do you see IBM and the work you're doing there, impacting this industry, and it's progress going forward?

Andy: That's really interesting because it's tempting to think that SCADA host is the be all and end all of the automation system. But as we talk a lot more about OT/IT integration. Sort of operational technologies and information technologies, so the SCADA world meets the business world of an enterprise, then the stack of applications that IBM has on top of the data, so in IBM cloud you've got things like, as part of the Watson Analytics and Watson AI family of products, we've got things like preventative maintenance. So we've got a product which enables you to link on top of an asset management system like Maximo, which tracks the run hours of a system out in the field, and then can start to predict when it's likely to fail. So rather than doing scheduled maintenance on a fixed interval, we can now get a prediction from Watson, which will say, that pump is likely to fail in 14 days time. So you can then do proactive maintenance, so it doesn't actually fail.

These insights that you can get into the data, are really what's enabled by the stack of software from the device down to bottom through Sparkplug, going to Ignition, going into the IBM Cloud Injector, going up to the Watson IT platform, going to IBM's cloud. Then you've got applications there which can give you the real insights, the business insights into the data, which really first of all cement the IT/OT integration, but also give the maintenance managers and the SCADA managers the insights they need to keep their equipment, their plants, their pipelines running day in day out.

Don: That's a really nice vision of the future, and certainly the role that IBM can play. I just want to say something in relationship to what we do here in Inductive Automation. That one of the challenges is to get people to think differently about how they architect, how they conceive. To think of the possible and yet they're very often in the world that we live in with our customers, they are running plants 24 by seven, and they can't take a three foot up, let alone 30 thousand feet. So what do you think could accelerate? I mean, we look at the Ignition platform, and we go, my gosh the capability to fit into something you just described is clearly we'd play a role in that. You play a role in that. Many people in the vendor community are doing things to play a role in that. So what do you think would maybe help accelerate adoption or major barriers to adoption of real IIoT solutions among these industrial organizations, so they can actually realize some of these benefits?

Andy: One of the most interesting panel discussions this week, which I was actually part of, was the conversation about open source versus proprietary protocols, and open source software. The discussion that came around that, the industrial sector has been very much tied into the proprietary world for many, many years. So that's just kind of the status quo of the momentum that's carried forward. But increasingly we're seeing open source systems, open protocols, open source software where people are collaborating around a piece of software, or a protocol, or a way of doing things that can be shared across the industry. Even people who compete with each other are prepared to share things down at the commodity level, for say that's the baseline, so if we've all got to do, let's just make it better for everyone.

So I think the answer to your question is really people working together as a community on problems to solve the common problems that face the industry, and then to use that as a base platform to then build their USP to differentiate themselves in the competitive landscape. It's what we're seeing with the ICC community, with Allen Ray heading that, shared the start of for example, people are sharing templates in the cloud. People sharing algorithms that they know solve particular problems. Just making that knowledge available so people didn't have to reinvent the wheel over and over again, which means all these businesses become more agile, more lean, able to deliver better service to their customers at a lower cost, which helps sees a virtuous course, which helps everybody. That collaboration and openness, which is a trend that's been in the last, sort of as we got into open source, over the last 15, 20 years has really been more and more apparent.

Those of us who work in open, agile environment, makes complete sense to be working in that way. So just going to get more and more thinking that way, and things like the community you built Inductive Automation, and Ignition platform really great ways to start doing that.

Don: So Andy, thanks. I really appreciate you taking the time to sit down, and share some thoughts today. Any final thoughts you may want to share with the audience? You know we have a diverse group of people listening to this. So what would you say that you want to leave them with?

Andy: So I think we're in such an interesting time now. The sophistication of systems, of systems of systems, of the building blocks that we have to build application with are becoming so complex and sophisticated that you can now build really amazing solutions really easily, particularly when you start putting stuff in the cloud, rather than writing tens of thousands of lines of code. You can just make a few calls to APIs out in the cloud, and link them together to make new applications. The consequence of that is that we're now seeing applications which can be cross-industry. So everyone should be looking out around them.

I know you said everyone's busy, and it's hard to raise your head above the parapet, but just the chance to look at other industries, see what they're doing, see how they're solving problems that you'll find will be analogous to problems you're trying to solve, means you'll suddenly get insights into ways of solving problems that you didn't get before. Quite often when I go and talk to my clients, the value I bring is that I can be talking to somebody in say the banking sector and say, "Oh, I know something in healthcare they're using to track assets, which I can see how you could use that to track something in your organization." Those cross fertilization of ideas that are so powerful.

So take the time to go to conferences, talk to people, look out across other industries. Maybe send scouting parties out to look for new technologies. Have brown bag lunches where if somebody comes in from outside, speaker to tell you about some new technology, because that's the way we're going to learn.

Everything is moving so fast. Disruption is the new normal. One thing you can be absolutely certain of is that it's going to keep changing so don't think, oh we'll just wait for the next big thing, then figure that out, and stay with that for the next 20 years, because you'll already be left behind as the next big thing is coming along. Constant churn. Constant agility. The need to be responsive and stay fresh in the competitive landscape is what everyone's got to do. So that's what I encourage people to be inspired to do.

Don P: Great final words. Andy Stanford Clark, thank you so much for being our guest today.

Andy: Thank you Don. It’s been a real pleasure.