Handling ‘zettabytes of data’ at the edge: Dave Laurello, CEO of Stratus Technologies
Image credit: Nick Smith
Dave Laurello, CEO of US-based Stratus Technologies, discusses the evolving trend for data collection, storage and analysis in the Industrial Internet of Things (IIoT) and explains how ‘edge computing’ is gaining traction in a world where data is ubiquitous.
“There’s a trend that is often referred to as the ‘digitalisation of our lives’ for the purpose of analysis,” says Dave Laurello, veteran computer and network design engineer, and long-standing chief executive of Stratus Technologies.
“What this means is there are smart devices out there collecting data. We tend to think of smart devices on a manufacturing line monitoring how a system is performing in, say, a wastewater plant, processing data 24/7. That data is being collected and analysed. A chemical is being made. A drug is being made. Every millisecond we are looking at the process to make sure the formula is correct.”
The estimates vary, but Laurello says that by 2020 there will be 15 billion smart devices connected to the Internet of Things (IoT). “What this means is there will be twice as many devices as there are people in the world. And they are collecting lots of data. By 2020 they’ll be collecting 11 zettabytes of data [1ZB = 1021 bytes].” Laurello explains that while it was once thought all of this data would be pushed up into the cloud for analysis, “the reality is if you want to make real-time changes close to the process line, there’s too much latency in this – it takes too long to get up to the cloud and back. There’s not enough bandwidth in the universe to take 11ZB of data up to the cloud and back again. There are also security issues, and then there is the cost of computing and transacting data in the cloud. People are starting to realise we should start to analyse data close to its capture. We’ll then send some data to the cloud for post-processing.” Laurello thinks that with the advent of ‘edge computing’, 40-60 per cent of generated data will be analysed and stored locally.
As the number of smart devices continues to grow “people are pushing applications down to the process line. You’re going to have real-time analysis and real-time artificial intelligence to the point where you will no longer need people in these locations. So the trend is that with a lot of data being generated at the edge, you need to do a lot of computing near the edge.” It follows that the growth of edge computers is, according to Laurello ‘dramatic’.
As the concept of the edge is relatively new, it’s not immediately obvious how to define it in precise terms. One generally evolving definition says that it is anything ‘outside the data centre’, which could be confusing, especially in the context of small data centres. Zoning in, Laurello says: “I like to think of the edge as computing close to the source of the data.” At this frontier, “just as there is a variety of computers in the data centre, there is a variety of computers at the edge. I don’t think you’ll see large data centre-like machines. But you’ll see servers that have varying capacities of CPU memory.”
Edge computing has its challenges. “For one thing, the computers themselves can be located in isolated areas,” says Laurello, “maybe along an oil and gas distribution line, a train track, a remote manufacturing facility. In these areas, there tend to be no IT resources.” The second challenge is security: “You have these computers in, say, a wastewater plant, and the reality is there maybe no one there for days.” The third is the criticality of the data. “You’re collecting this data to analyse it. If the computer goes down, you can’t do that analysis.”
This is where Stratus Technologies comes in. “We’ve created a vision statement that defines what attributes an edge computer should have. We think it should have three basic attributes: it should be simple to deploy and manage; it should be protected; and, it should be autonomous.
“‘Simple’ means it can be installed, managed and repaired by a non-IT person – so an OT resource can handle that when you’re trying to avoid putting an IT resource out in these locations. ‘Protected’ means it should be physically secure, so you can’t just walk up to it and put a USB stick into it, as well as having full cyber security to be able to make sure the network is protected. Of course, it’s got to be continuously available with 100 per cent uptime. Third, it needs to be ‘autonomous’, in that it is hands-free. You want to be able to remotely monitor that system to make sure it is performing, and if you need to diagnose something you can do that remotely without sending people out into the field. Eventually, you want to make it self-healing so it takes corrective action without the need for a person.”
Stratus has released its first edge-optimised product family – the ztC Edge (‘zero touch computing’), which directly corresponds to the three general attributes Laurello describes. “You can set it up in 30 minutes without having to be an IT person. When you open the box, you get a one-page assembly sheet that an eight-year-old can follow. Our first product was the ztC Edge 100i, which is two industrialised servers with cables and power plugs, just like connecting an Xbox to your TV. Our strategy is to use off-the-shelf hardware, while our intellectual capital is in the middle-ware and interface.”
Apart from serving for four years in the US military, Laurello has always been an engineer. He read engineering (BSEE) at the University of Massachusetts, “and then started working designing computers for a small memory test company that doesn’t exist any more and then a digital equipment company that doesn’t exist any more. I spent a lot of time in that space and fell in love with technology and building computers.”
As well as being heavily involved in design, Laurello recognised a deep satisfaction in “managing those efforts. After designing for a while, I got a lot of enjoyment out of managing a team, and then I moved into the business side and became a GM in a network company and kept moving up from there.” He joined Stratus in 2000, where he found the company had “a very 1980s-ish computer architecture style that was very proprietary. They made everything in big boxes and cost a million dollars each. My challenge was how to get that down to $50,000. I had a lot of kids with T-shirts and surf boards working for me and we went to town.”
Making the transition from engineer to engineering manager with oversight of a successful company, understanding markets and finance while staying in touch with the techies at the coalface, isn’t always an easy move. Yet for Laurello the transition was smooth, helped in no small part by the fact he has always believed the pathway to successful leadership is “to surround yourself with good people who know how to do things you don’t know how to do. If you can do that, and you have the ability to recruit talent, then that goes a long way to being a successful leader. There are other attributes you need to develop, but I think this is the key one. Sometimes if you pick people too much like you and have your exact same skillset, those teams won’t do so well. But if you’re not afraid to pick someone who’s smarter than you are, especially in areas you might not be up to speed on, you will end up having a very strong team and will create a diversity that you can build on.”
That said, Laurello cheerfully admits that being able to listen to people smarter than you is a “difficult skill to acquire. Learning to admit you can sometimes be wrong is difficult. Yet you’ve just got to get over it and move on.”
He also thinks his age and experience of four decades plays into his hands: “when you’re younger you have emotional peaks and valleys that are very high and very low. As you age, these mellow out and you’re able to gain a better perspective.”
‘Edge computers need to be simple to deploy, protected and autonomous.’
Responding to a world where there is expectation of continuous collection, storage and analysis of process data, the key management strategy for Laurello was to identify the emerging trend for edge computing. “My philosophy is that every company has a unique value proposition; something they do better than anybody else.” For Stratus, Laurello believes he makes “the most reliable systems on the planet, and we’ve done that for almost 40 years.” It’s a bold claim, but for the CEO who’s been with the company since the dawn of the millennium, if nothing else, it shows commitment.
“It’s going to sound familiar, but all of our work has been in the areas of making the product continuously available, operationally simple and autonomous. Four decades ago, Stratus’ product was put into telecommunications offices that were brick buildings with no people inside them. Since that time the world has evolved,” says the 63-year-old engineer. “But what I think has happened is that it’s almost ‘back to the future’ for us, in that changes in the market has made our value proposition, which was evolved in the 1980s, become even more relevant today. And that is edge computing.”
For the ‘time in the middle’, Laurello admits the company “just became better at what we were doing and working on lower price points. Yet edge computing gave the way we do things a second life. We started looking at where we were winning and we saw about three or four years ago that this was in industrial automation, which was growing at more than 20 per cent a year. As we looked deeper into this, we realised these were edge environments. At the same time, the word ‘edge’ was starting to be used and the market was beginning to evolve. It was an opportunity to put together our value proposition with the market dynamic and offer an ‘edge-optimised’ product.”
This change in market dynamic came about because of the blurring of the line between operational technology (OT) and IT. But Laurello prefers not to call this, as others do, a ‘convergence’. “I’d say it’s more a meeting of OT and IT. We think it’s more of a hybrid OT model in which OT individuals will become more IT-literate. Yet the reality is that when you go into small or medium enterprises such as a brewery, what these folks are worrying about is their business model. They don’t really want to become IT experts. They recognise the need for the process control and the monitoring, but they don’t want to hire IT resources. They would like the benefits of IT, but in a form they are set up to deal with.”
Laurello thinks the edge is far from a done deal. “We believe that in particular reference to the IIoT, this will evolve in phases. What we have seen as a first phase for organisations that have maybe had their process in place for 15 years, is they are probably running SCADA with lots of servers each with a unique application. They have what I would call ‘server sprawl’ and have patch issues. What we see these people doing is bringing virtualisation technology to the edge. That’s the first step: it’s not too intrusive and it’s not really affecting their process.” Step two, forecasts Laurello, will be in the form of “new applications such as real-time analytics and artificial intelligence. These are still in the incubation phase. Yet efficiency applications will come to the fore.”
As to the future, Laurello believes that when it comes to the age of digitalisation “we’re just getting started”, but he has a concern about how much of the ever-increasing zettabytes of data is ever going to get meaningfully used.
“I think we’re still struggling with technologies that take raw data and actually turns it into useable information. That is to come. Real-time artificial intelligence is still in its prototype phase, but you can imagine in 10 years’ time that facilities that once relied on hundreds of people will be run by maybe 10.
“Operational efficiency will improve, but it will create societal issues around the people who once did those jobs. The focus will be people-less manufacturing plants. When you look at manufacturers today, the biggest cost we’re left with is people. That’s the last realm of efficiency to deal with and we will definitely get to the place where these plants are purely automated.”