Featured interview - Steve Pavlosky

Steve Pavlosky is principal product manager of historian and data at the edge technology for GE Digital.

Steve Pavlosky, GE Digital

ADDEDWednesday, January 20, 2021

  • Q
    What makes GE Digital optimistic about the future of the automation industry?
    It’s a long-standing, almost constant growth market. Industrial customers are always faced with new challenges. Think about COVID. COVID has really accelerated change. The people that used to be on the floor aren’t necessarily on the floor anymore.
    Before some customers were leery about cloud technology. COVID has changed that faster than anything else in the last 5 years because customers have had to adapt. They’ve realized they could do that safely and securely. 
    This notion of technology changing around us, manufacturers continuing to need to improve productivity and safety and quality, that’s always going to be a challenge. It’s really a fun industry to be in because there’s always new challenges and technology. There’s always the ability to bring new solutions to the customers. 
    The last 5 or 6 years with this – whether you call it industry 4.0 or whether you call it industrial internet of things and the cloud and high-speed networking and edge computing, it has really been one of the most exciting times in my 35 years in the industry.
  • Q
    What is the biggest challenge that integrators are facing today and how is GE Digital helping?
    The pace of technology is as painful for integrators, perhaps more so than for the end users. Once the end user has made a decision, they bought a set of technology – OK, they get comfortable with it and they can use it for some period.
    With the integrators, technology is always changing. One of the things we’re trying to do within GE Digital is frequent communication with our integrator community, talking about product changes and relating how those product changes and enhancements can help them solve additional problems for customers. As they talk to their customers, what problems are the customers describing that the new technology could potentially solve?
    And then be educated enough to both recognize that and to be able to implement efficiently. Because that’s the other problem that integrators face is that time is money. 
    Learning is expensive for an integrator. Any hour spent reading a manual, watching YouTube videos, is time that’s not being applied to a billable hour to a customer. 
    That’s another piece of where GE Digital over time perhaps didn’t put enough focus on and that’s something that we’re actually focused very heavily on today, which is how do we take our products and make them much faster and easier to deploy in a way that we hadn’t really focused on perhaps a couple of years ago.
    So that should then help integrators be more profitable. So, they should be able to get more done in a day. We’re focused on thinking about cost of deployment in addition to cost of long-term management, which is really the burden the end user plays. The cost of deployment, which is what the system integrator faces.
  • Q
    What are the opportunities for integrators and end customers in edge computing?
    There remains a very large opportunity for analytics at the machine on edge devices that are smart enough and have the communications capability to also send data, the right data, for data storage.
    So, there are big opportunities there from an edge computing perspective to help those users get more productivity out of those machines.
  • Q
    What kinds of trends and challenges are you seeing in edge computing?
    Edge computing first, I mean there’s lots of definitions, right? There’s what we often think of as small footprint edge devices that fit in the IoT world where people are doing data collection off of machines and then sending that data either to the cloud or some application.
    From a trend perspective, more and more compute capacity, more and more network capacity and you know, when you think about what’s 5G going to do that in that space. I don’t have the answer to that. I don’t think anybody does. But I’ve certainly talked to customers who have been starting to look at how does 5G work. 
    The challenge is really the pace of change, which creates a problem for the industrial user and the management of devices. You think about Windows laptops have existed and yes, they get refreshed every 18 months.
    But IT can manage a fleet of Windows laptops because they’re all basically the same in their behavior. They’re all running the same operating system. 
    But the sort of variation and speed of change of edge devices presents a big challenge to the end user community today and to then the system integrators who are potentially recommending technology. 
  • Q
    What differentiates GE from the other competitors around historians?
    There is this notion that we come from an industrial background versus a software-only player. While there are some good solutions that come from software-only players, that’s one aspect of taking a look at GE.
    The real question to think about as they’re researching technology is not to think about their problem today but to think about their problem 3 years from now. 
    What might you want to be doing 3 years from now? What analytics? How might you want to tie that data together? How might you want to tie the data from this plant together with another plant? 
    We have Windows-based historian because most customers in the industrial space, they’re running Windows-based applications. But we imported our historian to Linux. We have some customers that are Linux shops, and they want to run on Linux.
    So does the technology platform work? What’s the sort of expertise and how is the vendor – what’s their longevity. GE has been in this business since the ‘70s when we were doing numerical control pre-processing. And then moved into control systems and HMI/SCADA and MES and the like.
    So, there are many different factors but the key when making a buying decision, is to think about how the technology is going to integrate. How does it fit the current infrastructure requirements? And think about a solution not from the needs of today but what the needs of 3 years from now or 2 years from now will be, so that they’re making a decision on a platform or a product that can scale to what they might need.
  • Q
    Help me understand when it would make sense to use a historian solution versus a relational database.
    Historians are purpose-built to store, retrieve and distribute primarily time-series data very efficiently. There can be other data types. There can be alarm-and-event data. There could be binary object data. There can be textual data. But in general, it’s sequenced over time versus relational databases which are really better at data sets where the data structure changes over time and people need to be able to flexibly change table layouts or build what they need.
    In the case of time series data storage, what needs to be collected is known, right? It’s the data value. At what time? What’s the quality? So, you can very efficiently store data, and while people will say, “Well, hard disk space is not expensive anymore. I don’t care about that.”
    To me it’s really about the retrieval. If you can very efficiently store data and very efficiently retrieve it, you’re not having people wait for that answer and, in addition to that, historians typically have other capabilities. They have built in, “How am I going to get that OT data into the system reliably? What happens if a network connection occurs?”
    All the players in the market have thought through storing forward-type technologies and aggregation technologies. You want to be able to easily collect the average of a value over an hour. Well, that’s built into a historian versus a relational database itself doesn’t have that. It’s just a storage mechanism.
    Now there are people that have built historian solutions on top of relational databases and those have their place. I don’t think that they are scalable as purpose-built solutions. But they do have their place. They’re kind of a hybrid.
  • Q
    What about more recently, like this past year and going forward into the next 12, 24 months? How has GE Digital grown and changed and what do you expect coming up?
    It has been an interesting period for GE Digital. GE Digital was formed not that long ago in the scope of GE. GE has been around over 100 years. GE Digital is maybe 5 or 6 years old. 
    At the outset, GE Digital was largely focused on the industrial internet, cloud-based technologies as sort of a platform player. As I think about what has happened over the last 2 years, we’ve really narrowed our focus to how do we enable customers to get value faster from technology by focusing on purpose-built applications – as opposed to being a core platform player.
    I don’t see any change in that strategy. What we really are starting to think about is continuing down the path of not only making it easier to collect data and get context to the data that’s being collected, but also analytics. A lot of our customers are asking about analytics. We’ve had data scientists as part of our organization, and we build analytics for our own purpose-built use cases to help turbines run better, to help compressors run better or to help the grid operate better or help customers understand what the important inputs into a are given process that impact the quality.
    What we’re really trying to do is now better understand common use cases of data, common problems to solve and sort of pre-packaging analytics so that customers can get value faster from the technology.
  • Q
    What are some of the changes that you’ve seen related to industrial data management over the years?
    There has obviously been a tremendous amount of technology change since I started in the mid-80s. I won’t spend a lot of time on that. But at the very beginning, PLCs and programmable controls for equipment were relatively new. HMI/SCADA systems were brand new. I was there at GE’s release of their first version of Simplicity in 1988-1989 timeframe. So, these are products that have lived for a long time in the market.
    But really, they were all about – even when you think about HMI/SCADA -- it’s about how do we create context for the data that’s coming from a machine so an operator can take action. That basic concept still lives on today.
    We may use different language, or we may use different technology. But it’s really about enabling humans by and large to take advantage of technology so that they can make better decisions around their operations and drive uptime, productivity, reduce cost, improve quality, all of those kind of business challenges.