How can global enterprise cybersecurity be improved for better enterprise integrity and risk mitigation? What constitutes a good standard, or set of standards, to help? And how can organizations work to better detect misdeeds, rather than have attackers on their networks for months before being discovered?
These questions were addressed during a February panel discussion at The Open Group San Diego 2015 conference. Led by moderator Dave Lounsbury, Chief Technology Officer, The Open Group, the speakers included Edna Conway, Chief Security Officer for Global Supply Chain, Cisco; Mary Ann Mezzapelle, Americas CTO for Enterprise Security Services, HP; Jim Hietala, Vice President of Security for The Open Group, and Rance DeLong, Researcher into Security and High Assurance Systems, Santa Clara University.
Here are some excerpts:
Dave Lounsbury: We’ve heard about the security, cybersecurity landscape, and, of course, everyone knows about all the many recent breaches. Obviously, the challenge is growing in cybersecurity. So, I want to start asking a few questions, directing the first one to Edna Conway.
We’ve heard about the Verizon Data Breach Investigation of DBIR report that catalogs the various attacks that have been made over the past year. One of the interesting findings was that in some of these breaches, the attackers were on the networks for months before being discovered.
What do we need to start doing differently to secure our enterprises?
Edna Conway: There are a couple of things. From my perspective, continuous monitoring is absolutely essential. People don’t like it because it requires rigor, consistency, and process. The real question is, what do you continuously monitor?
It’s what you monitor that makes a difference. Access control and authentication, should absolutely be on our radar screen, but I think the real ticket is behavior. What kind of behavior do you see authorized personnel engaging in that should send up as an alert? That’s a trend that we need to embrace more.
The second thing that we need to do differently is drive detection and containment. I think we try to do that, but we need to become more rigorous in it. Some of that rigor is around things like, are we actually doing advanced malware protection, rather than just detection?
What are we doing specifically around threat analytics and the feeds that come to us: how we absorb them, how we mine them, and how we consolidate them?
The third thing for me is how we get it right. I call that team the puzzle solvers. How do we get them together swiftly?
How do you put the right group of experts together when you see a behavior aberration or you get a threat feed that says that you need to address this now? When we see a threat injection, are we actually acting on the anomaly before it makes its way further along in the cycle?
Mary Ann Mezzapelle: Another thing that I’d like to add is making sure you have the executive support and processes in place. If you think how many plans and tests and other things that organizations have gone through for business continuity and recovery, you have to think about that incident response. We talked earlier about how to get the C suite involved. We need to have that executive sponsorship and understanding, and that means it’s connected to all the other parts of the enterprise.
So it might be the communications, it might be legal, it might be other things, but knowing how to do that and being able to respond to it quickly is also very important.
Rance DeLong: I agree on the monitoring being very important as well as the question of what to monitor. There are advances being made through research in this area, both modeling behavior — what are the nominal behaviors — and how we can allow for certain variations in the behavior and still not have too many false positives or too many false negatives.
Also on a technical level, we can analyze systems for certain invariants, and these can be very subtle and complicated invariance formulas that may be pages long and hold on the system during its normal operation. A monitor can be monitoring both for invariance, these static things, but they can also be monitoring for changes that are supposed to occur and whether those are occurring the way they’re supposed to.
Jim Hietala: The only thing I would add is that I think it’s about understanding where you really have risk and being able to measure how much risk is present in your given situation.
In the security industry, there has been a shift in mindset away from figuring that we can actually prevent every bad thing from happening towards really understanding where people may have gotten into the system. What are those markers that something is gone awry and reacting to that in a more timely way — so detective controls, as opposed to purely preventative type controls.
Lounsbury: We heard from Dawn Meyerriecks earlier about the convergence of virtual and physical and how that changes the risk management game. And we heard from Mary Ann Davidson about how she is definitely not going to connect her house to the Internet.
So this brings new potential risks and security management concerns. What do you see as the big Internet of Things (IoT) security concerns and how does the technology industry assess and respond to those?
Hietala: In terms of IoT, the thing that concern me is that many of the things that we’ve solved at some level in IT hardware, software, and systems seemed to have been forgotten by many of the IoT device manufacturers.
We have pretty well thought out processes for how we identify assets, we patch things, and we deal with security events and vulnerabilities that happen. The idea that, particularly on the consumer class of IoT type devices, we have devices out there with IP interfaces on them, and many of the manufacturers just haven’t had a thought of how they are going to patch something in the field, I think should scare us all to some degree.
Maybe it is, as Mary Ann mentioned, the idea that there are certain systemic risks that are out there that we just have to sort of nod our head and say that that’s the way it is. But certainly around really critical kinds of IoT applications, we need to take what we’ve learned in the last ten years and apply it to this new class of devices.
New architectural approach
DeLong: I’d like to add to that. We need a new architectural approach for IoT that will help to mitigate the systemic risks. And echoing the concerns expressed by Mary Ann a few minutes ago, in 2014, Europol, which is an organization that tracks criminal risks of various kinds, predicted by the end of 2014, murder by Internet, in the context of Internet of Things. It didn’t happen, but they predicted it, and I think it’s not farfetched that we may see it over time.
Lounsbury: What do we really know actually? Edna, do you have any reaction on that one?
Conway: Murder by Internet. That’s the question you gave me, thanks. Welcome to being a former prosecutor. The answer is on their derrieres. The reality is do we have any evidentiary reality to be able to prove that?
I think the challenge is one that’s really well-taken, which is we are probably all in agreement on, the convergence of these devices. We saw the convergence of IT and OT and we haven’t fixed that yet.
We are now moving with IoT into a scalability of the nature and volume of devices. To me, the real challenge will be to come up with new ways of deploying telemetry to allow us to see all the little crevices and corners of the Internet of Things, so that we can identify risks in the same way that we have. We haven’t mastered 100 percent, but we’ve certainly tackled predominately across the computer networks and the network itself and IT. We’re just not there with IoT.
Mezzapelle: Edna, it also brings to mind another thing — we need to take advantage of the technology itself. So as the data gets democratized, meaning it’s going to be everywhere — the velocity, volume, and so forth — we need to make sure that those devices can maybe be self-defendable, or maybe they can join together and defend themselves against other things.
So we can’t just apply the old-world thinking of being able to know everything and control everything, but to embed some of those kinds of characteristics in the systems, devices, and sensors themselves.
Lounsbury: We’ve heard about the need. In fact, Ron Ross mentioned the need for increased public-private cooperation to address the cybersecurity threat. Ron, I would urge you to think about including voluntary consensus standards organizations in that essential partnership you mentioned to make sure that you get that high level of engagement, but of course, this is a broad concern to everybody.
President Obama has made a call for legislation on enabling cybersecurity and information sharing, and one of the points within that was shaping a cyber savvy workforce and many other parts of public-private information sharing.
So what more can be done to enable effective public-private cooperation on this and what steps can we, as a consensus organization, take to actually help make that happen? Mary Ann, do you want to tackle that one and see where it goes?
Collaboration is important
Mezzapelle: To your point, collaboration is important and it’s not just about the public and the private partnership. It also means within an industry sector or in your supply chain and third-party. It’s not just about the technology; it’s also about the processes, and being able to communicate effectively, almost at machine speed, in those areas.
So you think about the people, the processes, and the technology, I don’t think it’s going to be solved by government. I think I agree with the previous speakers when they were talking about how it needs to be more hand-in-hand.
There are some ways that industry can actually lead that. We have some examples, for instance what we are doing with the Healthcare Forum and with the Mining and Minerals Forum. That might seem like a little bit, but it’s that little bit that helps, that brings it together to make it easier for that connection.
It’s also important to think about, especially with the class of services and products that are available as a service, another measure of collaboration. Maybe you, as a security organization, determine that your capabilities can’t keep up with the bad guys, because they have more money, more time, more opportunity to take advantage, either from a financial perspective or maybe even from a competitive perspective, for your intellectual property.
You really can’t do it yourself. You need those product vendors or you might need a services vendor to really be able to fill in the gaps, so that you can have that kind of thing on demand. So I would encourage you to think about that kind of collaboration through partnerships in your whole ecosystem.
DeLong: I know that people in the commercial world don’t like a lot of regulation, but I think government can provide certain minimal standards that must be met to raise the floor. Not that companies won’t exceed these and use that as a competitive basis, but if minimum is set in regulations, then this will raise the whole level of discourse.
Conway: We could probably debate over a really big bottle of wine whether it’s regulation or whether it’s collaboration. I agree with Mary Ann. I think we need to sit down and ask what are the biggest challenges that we have and take bold, hairy steps to pull together as an industry? And that includes government and academia as partners.
But I will give you just one example: ECIDs. They are out there and some are on semiconductor devices. There are some semiconductor companies that already use them, and there are some that don’t.
A simple concept would be if we could make sure that those were actually published on an access control base, so that we could go and see whether the ECID was actually utilized, number one.
Speeding up standards
Lounsbury: Okay, thanks. Jim, I think this next question is about standards evolution. So we’re going to send it to someone from a standards organization.
The cyber security threat evolves quickly, and protection mechanisms evolve along with them. It’s the old attacker-defender arms race. Standards take time to develop, particularly if you use a consensus process. How do we change the dynamic? How do we make sure that the standards are keeping up with the evolving threat picture? And what more can be done to speed that up and keep it fresh?
Hietala: I’ll go back to a series of workshops that we did in the fall around the topic of security automation. In terms of The Open Group’s perspective, standards development works best when you have a strong customer voice expressed around the pain points, requirements, and issues.
We did a series of workshops on the topic of security automation with customer organizations. We had maybe a couple of hundred inputs over the course of four workshops, three physical events, and one that we did on the web. We collected that data, and then are bringing it to the vendors and putting some context around a really critical area, which is how do you automate some of the security capabilities so that you are responding faster to attacks and threats.
Generally, with just the idea that we bring customers into the discussion early, we make sure that their issues are well-understood. That helps motivate the vendor community to get serious about doing things more quickly.
One of the things we heard pretty clearly in terms of requirements was that multi-vendor interoperability between security components is pretty critical in that world. It’s a multi-vendor world that most of the customers are living with. So building interfaces that are open, where you have got interoperability between vendors, is a really key thing.
DeLong: It’s a really challenging problem, because in emerging technologies, where you want to encourage and you depend upon innovation, it’s hard to establish a standard. It’s still emerging. You don’t know what’s going to be a good standard. So you hold off and you wait and then you start to get innovation, you get divergence, and then bringing it back together ultimately takes more energy.
Lounsbury: Rance, since you have got the microphone, how much of the current cybersecurity situation is attributed to poor blocking and tackling in terms of the basics, like doing security architecture or even having a method to do security architecture, things like risk management, which of course Jim and the Security Forum have been looking into? And not only that, what about translating that theory into operational practice and making sure that people are doing it on a regular basis?
DeLong: A report I read on SANs, a US Government issued report on January 28 of this year, said that that many, or most, or all of our critical weapons systems contain flaws and vulnerabilities. One of the main conclusions was that, in many cases, it was due to not taking care of the basics — the proper administration of systems, the proper application of repairs, patches, vulnerability fixes, and so on. So we need to be able to do it in critical systems as well as on desktops.
Mezzapelle: You might consider the open-source code crisis that happened over the past year with Heartbleed, where the benefits of having open-source code is somewhat offset by the disadvantages.
That may be one of the areas where the basics need to be looked at. It’s also because those systems were created in an environment when the threats were at an entirely different level. That’s a reminder that we need to look to that in our own organization.
Another thing is in mobile applications, where we have such a rush to get out features, revs, and everything like that, that it’s not entirety embedded in the system’s lifecycle or in a new startup company. Those are the some of the other basic areas where we find that the basics, the foundation, needs to be solidified to really help enhance the security in those areas.
Hietala: So in the world of security, it can be a little bit opaque, when you look at a given breach, as to what really happened, what failed, and so on. But enough information has come out about some of the breaches that you get some visibility into what went wrong.
Of the two big insider breaches — WikiLeaks and then Snowden — in both cases, there were fairly fundamental security controls that should have been in place, or maybe were in place, but were poorly performed, that contributed to those — access control type things, authorization, and so on.
Even in some of the large retailer credit card breaches, you can point to the fact that they didn’t do certain things right in terms of the basic blocking and tackling.
There’s a whole lot of security technology out there, a whole lot of security controls that you can look to, but implementing the right ones for your situation, given the risk that you have and then operating them effectively, is an ongoing challenge for most companies.
Mezzapelle: Can I pose a question? It’s one of my premises that sometimes compliance and regulation makes companies do things in the wrong areas to the point where they have a less secure system. What do you think about that and how that impacts the blocking and tackling?
Hietala: That has probably been true for, say, the four years preceding this, but there was a study just recently — I couldn’t tell you who it was from — but it basically flipped that. For the last five years or so, compliance has always been at the top of the list of drivers for information security spend in projects and so forth, but it has dropped down considerably, because of all these high profile breaches. Senior executive teams are saying, “Okay, enough. I don’t care what the compliance regulations say, we’re going to do the things we need to do to secure our environment.” Nobody wants to be the next Sony.
Mezzapelle: Or the Target CEO who had to step down. Even though they were compliant, they still had a breach, which unfortunately, is probably an opportunity at almost every enterprise and agency that’s out there.
The right eyeballs
DeLong: And on the subject of open source, it’s frequently given as a justification or a benefit of open source that it will be more secure because there are millions of eyeballs looking at it. It’s not millions of eyeballs, but the right eyeballs looking at it, the ones who can discern that there are security problems.
It’s not necessarily the case that open source is going to be more secure, because it can be viewed by millions of eyeballs. You can have proprietary software that has just as much, or more, attention from the right eyeballs as open source.
Mezzapelle: There are also those million eyeballs out there trying to make money on exploiting it before it does get patched — the new market economy.
Lounsbury: I was just going to mention that we’re now seeing that some large companies are paying those millions of eyeballs to go look for vulnerabilities, strangely enough, which they always find in other people’s code, not their own.
Mezzapelle: Our Zero Day Initiative, that was part of the business model, is to pay people to find things that we can implement into our own products first, but it also made it available to other companies and vendors so that they could fix it before it became public knowledge.
Some of the economics are changing too. They’re trying to get the white hatter, so to speak, to look at other parts that are maybe more critical, like what came up with Heartbleed.
Lounsbury: On that point, and I’m going to inject a question of my own if I may, on balance, is the open sharing of information of things like vulnerability analysis helping move us forward, and can we do more of it, or do we need to channel it in other ways?
Mezzapelle: We need to do more of it. It’s beneficial. We still have conclaves of secretness saying that you can give this information to this group of people, but not this group of people, and it’s very hard.
In my organization, which is global, I had to look at every last little detail to say, “Can I share it with someone who is a foreigner, or someone who is in my organization, but not in my organization?” It was really hard to try to figure out how we could use that information more effectively. If we can get it more automated to where it doesn’t have to be the good old network talking to someone else, or an email, or something like that, it’s more beneficial.
And it’s not just the vulnerabilities. It’s also looking more towards threat intelligence. You see a lot of investment, if you look at the details behind some of the investments in In-Q-Tel, for instance, about looking at data in a whole different way.
So we’re emphasizing data, both in analytics as well as threat prediction, being able to know where some thing is going to come over the hill and you can secure your enterprise or your applications or systems more effectively against it.
Lounsbury: Let’s go down the row. Edna, what are your thoughts on more open sharing?
Conway: We need to do more of it, but we need to do it in a controlled environment.
We can get ahead of the curve with not just predictive analysis, but telemetry, to feed the predictive analysis, and that’s not going to happen because a government regulation mandates that we report somewhere.
So if you look, for example, DFARS, that came out last year with regard to concerns about counterfeit mitigation and detection in COTS ICT, the reality is not everybody is a member of GIDEP, and many of us actually share our information faster than it gets into GIDEP and more comprehensively.
I will go back to it’s rigor in the industry and sharing in a controlled environment.
Lounsbury: Jim, thoughts on open sharing?
Hietala: Good idea. It gets a little murky when you’re looking at zero-day vulnerabilities. There is a whole black market that has developed around those things, where nations are to some degree hoarding them, paying a lot of money to get them, to use them in cyberwar type activities.
There’s a great book out now called ‘Zero Day’ by Kim Zetter, a writer from Wired. It gets into the history of Stuxnet and how it was discovered, and Symantec, and I forget the other security researcher firm that found it. There were a number of zero-day vulnerabilities there that were used in an offensive cyberwar a capacity. So it’s definitely a gray area at this point.
DeLong: I agree with what Edna said about the parameters of the controlled environment, the controlled way in which it’s done. Without naming any names, recently there were some feathers flying over a security research organization establishing some practices concerning a 60- or 90-day timeframe, in which they would notify a vendor of vulnerabilities, giving them an opportunity to issue a patch. In one instance recently, when that time expired and they released it, the vendor was rather upset because the patch had not been issued yet. So what are reasonable parameters of this controlled environment?
Lounsbury: Let’s move on here. Edna, one of the great quotes that came out of the early days of OTTF was that only God creates something from nothing and everybody else is on somebody’s supply chain. I love that quote.
But given that all IT components, or all IT products, are built from hardware and software components, which are sourced globally, what do we do to mitigate the specific risks resulting from malware and counterfeit parts being inserted in the supply chain? How do you make sure that the work to do that is reflected in creating preference for vendors who put that effort into it?
Conway: It’s probably three-dimensional. The first part is understanding what your problem is. If you go back to what we heard Mary Ann Davidson talk about earlier today, the reality is what is the problem you’re trying to solve?
I’ll just use the Trusted Technology Provider Standard as an example of that. Narrowing down what the problem is, where the problem is located, helps you, number one.
Then, you have to attack it from all dimensions. We have a tendency to think about cyber in isolation from the physical, and the physical in isolation from the cyber, and then the logical. For those of us who live in OT or supply chain, we have to have processes that drive this. If those three don’t converge and map together, we’ll fail, because there will be gaps, inevitable gaps.
For me, it’s identifying what your true problem is and then taking a three-dimensional approach to make sure that you always have security technology, the combination of the physical security, and then the logical processes to interlock and try to drive a mitigation scheme that will never reduce you to zero, but will identify things.
Particularly think about IoT in a manufacturing environment with the right sensor at the right time and telemetry around human behavior. All of a sudden, you’re going to know things before they get to a stage in that supply chain or product lifecycle where they can become devastating in their scope of problem.
DeLong: As one data point, there was a lot of concern over chips fabricated in various parts of the world being used in national security systems. And in 2008, DARPA initiated a program called TRUST, which had a very challenging objective for coming up with methods by which these chips could be validated after manufacture.
Just as one example of the outcome of that, under the IRIS Program in 2010, SRI unveiled an infrared laser microscope that could examine the chips at the nanometer level, both for construction, functionality, and their likely lifetime — how long they would last before they failed.
Lounsbury: Jim, Mary Ann, reactions.
Finding the real problem
Mezzapelle: The only other thing I wanted to add to Edna’s comment was reiteration about the economics of it and finding where the real problem is. Especially in the security area, information technology security, we tend to get so focused on trying to make it technically pure, avoiding the most 100 percent, ultimate risk. Sometimes, we forget to put our business ears on and think about what that really means for the business? Is it keeping them from innovating quickly, adapting to new markets, perhaps getting into a new global environment?
We have to make sure we look back at the business imperatives and make sure that we have metrics all along the road that help us make sure we are putting the investments in the right area, because security is really a risk balance, which I know Jim has a whole lot more to talk about.
Hietala: The one thing I would add to this conversation is that we have sort of been on a journey to where doing a better job of security is a good thing. The question is when is it going to become a differentiator for your product and service in the market. For me personally, a bank that really gets online banking and security right is a differentiator to me as a consumer.
I saw a study that was quoted this week at the World Economic Forum that said that, by 2:1 margin, consumers — and they surveyed consumers in 27 countries — think that governments and businesses are not paying enough attention to digital security.
So maybe that’s a mindset shift that’s occurring as a result of how bad cybersecurity has been. Maybe we’ll get to the point soon where it can be a differentiator for companies in the business-to-business context and a business-to-consumer context and so forth. So we can hope.
Conway: Great point. And just to pivot on that and point out how important it is. I know that what we are seeing now, and it’s a trend, and there are some cutting-edge folks who have been doing it for a while, but most boards of directors are looking at creating a digital advisory board for their company. They’re recognizing the pervasiveness of digital risk as its own risk that sometimes it reports up to the audit committee.
I’ve seen at least 20 or 30 in the last three months come around, asking, did you advise every board members to focus on this from multiple disciplines? If we get that right, it might allow us that opportunity to share the information more broadly.
Lounsbury: That’s a really interesting point, the point about multiple disciplines. The next question is unfortunately the final question — or fortunately, since it will get you to lunch. I am going to start off with Rance.
At some point, the difference between a security vulnerability failure or other kind of failures all flow into that big risk analysis that a digital-risk management regime would find out. One of the things that’s going on across the Real-Time and Embedded Systems Forum is to look at how we architect systems for higher levels of assurance, not just security vulnerabilities, but other kinds of failures as well.
The question I will ask here is, if a system fails its service-level agreement (SLA) for whatever reason, whether it’s security or some other kind of vulnerability, is that a result of our ability to do system architecture or software created without provably secure or provably assured components or the ability of the system to react to those kind of failures? If you believe that, how do we change it? How do we accelerate the adoption of better practices in order to mitigate the whole spectrum of risk of failure of the digital enterprise?
Emphasis on protection
DeLong: Well, in high assurance systems, obviously we still treat them as very important detection of problems when they occur, recovery from problems, but we put a greater emphasis on prevention, and we try to put greater effort into prevention.
You mentioned provably secure components, but provable security is only part of the picture. When you do prove, you prove a theorem, and in a reasonable system, a system of reasonable complexity, there isn’t just one theorem. There are tens, hundreds, or even thousands of theorems that are proved to establish certain properties in the system.
It has to do with proofs of the various parts, proofs of how the parts combine, what are the claims we want to make for the system, how do the proofs provide evidence that the claims are justified, and what kind of argumentation do we use based on that set of evidence.
So we’re looking at not just the proofs as little gems, if you will. A proof of a theorem think of it as a gemstone, but how are they all combined into creating a system?
If a movie star walked out on the red carpet with a little burlap sack around her neck full of a handful of gemstones, we wouldn’t be as impressed as we are when we see a beautiful necklace that’s been done by a real master, who has taken tens or hundreds of stones and combined them in a very pleasing and beautiful way.
And so we have to put as much attention, not just on the individual gemstones, which admittedly are created with very pure materials and under great pressure, but also how they are combined into a work that meets the purpose.
And so we have assurance cases, we have compositional reasoning, and other things that have to come into play. It’s not just about the provable components and it’s a mistake that is sometimes made to just focus on the proof.
Remember, proof is really just a degree of demonstration, and we always want some demonstration to have confidence in the system, and proof is just an extreme degree of demonstration.
Mezzapelle: I think I would summarize it by embedding security early and often, and don’t depend on it 100 percent. That means you have to make your systems, your processes and your people resilient.
This has been a special BriefingsDirect presentation and panel discussion from The Open Group San Diego 2015. Download a copy of the transcript. This follows an earlier discussion on cybersecurity standards for safer supply chains. Another earlier discussion from the event focused on synergies among major Enterprise Architecture frameworks. And a presentation by John Zachman, founder of the Zachman Framework.
Copyright The Open Group and Interarbor Solutions, LLC, 2005-2015. All rights reserved.
You may also be interested in: