When they start the process of figuring what to do, they start to build a plan.
“I think we are seeing a tipping point where people are actually starting to develop cybersecurity plans where they may not have had them before,” said Eric Knapp, senior fellow, chief engineer and director of cybersecurity research at Honeywell, during his interview with ISSSource Editor and Founder, Gregory Hale at the S4x23 ICS Security Conference in South Beach, Florida where they had a wide-ranging discussion on how to understand and reduce cyber risk.
ISSSource: Where do you see security awareness today compared to two or three years ago? When we started 13 years ago, awareness was number four or five on the list, but today it’s number one. Along with that, are you seeing organizations actually starting security programs?
Eric Knapp: That’s an interesting question because, we’ve both been doing this a long time, and back when we started, there was very little awareness and then there was awareness, but skepticism.Now, I think there’s a lot more general awareness, but there’s still a lack of specific awareness. Awareness of what they actually need to do. There’s an acknowledgment that something needs to be done, but a lot of people are still unsure what they actually have to do. Companies are going through the process of figuring that out. As they do that, they start to build the plan. I think we are seeing a tipping point where people are actually starting to develop cybersecurity plans where they may not have had them before.
‘I think we are seeing a tipping point where people are actually starting to develop cybersecurity plans where they may not have had them before.’
— Eric Knapp
ISSSource: With an increase in connectivity brought on by digitalization, how can a manufacturer go about securing that environment?
Eric Knapp: We developed a slide that we use a lot called the security journey. It’s basically a linear timeline of different cybersecurity controls, and services and where they fall on the roadmap of security maturity.
When you look at it in a line, doing an initial basic assessment at the very, very beginning if you don’t know anything is a great start. But there’s several other types of assessments you could do all the way up to a full-fledged risk assessment. Then even up to a HazOps type of a security system assessment.
It’s basically taking a risk assessment even further and factoring in the potential physical impact and physical countermeasures and safety controls, and all that. It’s having a super advanced risk assessment if you will. But if you haven’t started, you can’t jump straight there; it’s too much. There are many people here (at S4x23) trying to sell little shiny objects. A lot of those make sense, but they have a place along that journey as well.
Network anomaly detection is a great example because it’s so popular. There are all these companies, and they’re providing something that’s valuable, but that’s a bit more advanced. You can’t really take advantage of network anomaly detection until you have network segmentation, which a lot of companies don’t have.
You also need some sort of a team. When you get into monitoring, you start getting data. As soon as you start getting data, how do you look at it? Having a machine that analyzes and creates an alert doesn’t mean anything if you don’t have people that know what to look for and to do the right thing when they see it.
So, there has to be some incident response training in there. Everything fits. It’s fuzzy because some people can jump around, and some people have stronger needs in one way or another. But there is a fairly logical progression of what you need.
ISSSource: Before you understand what you need, you have to understand what you have.
Eric Knapp: That’s true, too. Some people don’t. Again, that goes to a basic assessment. One of the issues with the media blitz all of those network security companies made over the past few years is that they are selling an advanced control.
But then their primary selling point was, we’re going to discover your assets, which is a very, very basic need. Companies want to know what assets they have on their networks, but technology doesn’t always solve that basic need. Typically, a network discovery tool is going to find the stuff that, you know what it looks like. It’s not going to find the stuff like the off-brand PLC somebody bought on eBay because they had a project and then they forgot it was there – that’s not going to get detected. Nobody knows what that is. Who knows? It could be anything. So, you really need to – especially in the beginning – have to have people engaged.
You have to have people crawling under desks and following cables and, looking through closets. If wireless is involved, that makes it even harder, because you have to do things like spectral analysis and all that crazy stuff.
That stuff all has to get done before you can start relying on technology to help you attenuate. That’s the thing that got missed, I think, over the years. Because those companies did sell that, as a promise. “Oh, well, you just push a button, and we’ll find your whole network for you.”
We need to be able to discover the assets. Sure. And there are tools to do that when customers are ready for that, but most of the industry is still at the beginning. They really are. They have flat networks, that still have Internet-facing connections. If they have a firewall, it might not be implemented correctly.
Having said all that, there’s some interesting advancements happening right now. There’s a big surge in micro segmentation. I like those types of technologies because they help people do the stuff at the beginning of that journey.
How do you segment your network? You can do it the traditional way with switches and routers, which involves unplugging things, re-cabling stuff, running new cable in a lot of cases or implementing something wireless and reconfiguring routers. All of that is complex and disruptive.
Micro segmentation makes that easier. It’s taking something they need to do that’s hard and making it easier. That’s what this industry needs. We need to make it easier for everyone to do the basics. Because as soon as they do the basics, they can move on to the more advanced things and we need to move on, because the world is getting scary.
‘You also need some sort of a team. When you get into monitoring, you start getting data. As soon as you start getting data, how do you look at it? Having a machine that analyzes and creates an alert doesn’t mean anything if you don’t have people that know what to look for and to do the right thing when they see it.’
— Eric Knapp
ISSSource: In trying to reduce risk, does it vary from sector to sector, say from oil to utilities to autos to button making? Does it vary or is it pretty much the same thing across the board?
Eric Knapp: I would say it’s pretty much the same thing, but it does vary a little bit. If you dig into risk management methodologies –and make sure you have coffee with you when you do that – because it’s not a super entertaining topic. Risk is a factor of likelihood versus consequence.
The likelihood is you can start breaking down into those technical reasons why it might be likely like unpatched vulnerabilities, known vulnerabilities and active exploits out there. That makes it highly likely technically. Things like geopolitical likelihood needs to be considered also. If you’re in Ukraine right now, the likelihood that you’ll be attacked is a lot higher just because of that.
Then the impact is what happens if the attack is successful? Together, that is what makes risk.
If you were a nuclear facility and you’re talking about the reactor, the consequence of an incident is obviously huge. The likelihood probably less so, because that industry is a little bit further along on that maturity model than most, and has been for a long time now.
At a paper mill, the impact of an incident isn’t as catastrophic, but it’s still going to affect their ability to produce goods.
But there might be additional impacts that happen depending on where you are. If you operate a toxic chemical facility and it explodes, then you may not be able to sell your product, but there’s also going to be other consequences: reputation damage, and there could be a loss of life, contaminated facilities, and lawsuits to name a few.
Some continuous manufacturing in metals and plastics, if you have a hot molten product going through pipes and into molds and things, and that system shuts down when it’s not supposed to, you could potentially destroy the whole thing. Now, all your pipes are full of metal that isn’t molten.
But ultimately, when you go to a high enough level, it’s basically all the same thing: Protecting your industrial control processes.
ISSSource: Do you ever see a time where there won’t be any cyber threat hanging over the industry?
Eric Knapp: I don’t think so. A large part of the cyber threat is a commercial enterprise. It’s a business and it’s a business that makes money. So why would they stop? Or it’s a Nation State that is acting according to geo-political agendas.
ISSSource: Some industry experts say ransomware is the number one fear out there for this industry. Do you agree?
Eric Knapp: It’s probably the number one concern because of the organizations; again, back to the awareness question. It’s gotten to the C-level and the C-levels are worried about money. Ransomware can have a huge financial impact. Colonial Pipeline was the perfect example. It didn’t attack the pipeline, but it took the pipelines offline. Not only do they have to pay the ransom. But how much money did they lose?
Well, that product wasn’t being moved up and down the East Coast. It’s money they’ll never see again. From a corporate perspective, ransomware is the biggest concern. I’m differentiating between concern and fear. Again, I deal with threat research, and I put myself in those worst-case scenarios. I’m like, “What would make me afraid?” If price of gas goes up a dollar, I’m not going to be afraid. I’m going to be unhappy, but I’m not going to be afraid. But if a weaponized attack took part of the grid out, oh, I would be afraid.
ISSSource: Are your customers further along in their journey to have some kind of resiliency plan in place to know what to do if and when an attack occurs, so they can stay up and running?
Eric Knapp: Some of them are. There are certain customers who’ve been on that journey for a while. What tends to happen is, they’ll either start at the very beginning and say, “I know we need to start, so we’re going to start super basic.”
Or there was some sort of an incident and so they’re like, “Well, we have to start because this happened, and we may want to jump ahead and put a control in place to prevent this specific thing from happening again.” Sort of reactive at first, and then they’re able to step back and start on the journey a little less panicked. But once they’ve been doing that for a year or two, then they start to develop that real plan and they come back to us saying, “We want to do this. We want to do this. We want to do this.” It’s like a hockey stick. It starts slow, but it builds momentum.
In that journey, having resilience built in is really important. Not just on the security control side, but on the process control side.
One thing I will say is it’s a more advanced thing to deal with, but something people should deal with from early on is having an incident response plan. What happens when I get attacked? That could evolve over the years as you become a little bit more mature.
But there’s no guarantee you’re not going to get hit with an attack right now. “Okay, I’m going to start. I’m going to get a nice plan in place.” Then you find out, oh no, I just got attacked. Or it could come years later. You need to know what to do to minimize the impact of that when it happens.
It puts you in that mindset of how do I make the whole system more resilient? Because you’re starting thinking, “Once it’s broken, how do I put it back together?” But a big part of that for OT is recovery. How do I get my process up and running?
ISSSource: How does a resiliency plan work with a business continuity? Does one flow into the other, or are they two separate things?
Eric Knapp: Just like safety and security should flow into one another. In an ideal world, they would, and they should, but they probably don’t. Not so much.
Safety systems are extremely well thought out to prevent something that’s not safe from happening. But they approach it from physics. If there’s too much pressure, something will explode. Because in cybersecurity you’re not dealing with physics, you’re dealing with a manipulative, clever mind. If safety and security people get together and learn from each other, then you start building safety systems against things that shouldn’t happen, but could if somebody wanted them to. That’s where you get that cyber resiliency piece.
ISSSource: In the keynotes here at S4x23, they were talking about thinking outside the box and learning from other sectors. Is that something security people are doing, or are they sticking with the tried-and-true methods?
Eric Knapp: I see people doing that. For cybersecurity people, whether they have a job in a plant or whether they have a job doing research, the job trains us to think that way. Just like somebody who’s a safety person is trained to think along different lines. A safety threat model is what are all the possible things that could happen to control systems. A cyber threat model is what could happen that we don’t expect from outside our control system. It’s the same thought process, but it’s completely flipped.
We have to think about the things that we haven’t thought about. We have to try to come up with that. That’s the fun part of my job, as we get to try to develop new attacks. How could we break something in a different way that nobody’s ever tried to break it before, so we can figure out how to protect against it?
ISSSource: In the end, I like to leave readers with an idea of best practices. So, along those lines, what are the top three things the manufacturer can do to reduce cyber risk?
Eric Knapp: If we go right back to the risk equation, likelihood and consequence, we should work to minimize both of those things. How do you minimize the impact if something goes wrong? That’s where robust infrastructure and maybe a little bit out-of-the-box thinking on the safety side comes in. If an attack comes through, the consequence just isn’t that bad.
That’s one. The other one would be about making it less likely. That’s following all the best practices and having a secure infrastructure that is segmented properly.
I would say the third one is knowing what to do when it goes wrong anyway. If the impact is less and the likelihood is less – and it’s easier to recover when it happens – then wouldn’t it be great if you got that call on Friday night and was like, “Oh my God, we just got attacked.” You’d be like, “Oh, honey, I’ll be half an hour late because there’s this irritating thing that happened.” We’ll be able to take care of it in a second because we know what we’re doing.

