This blog was written by Wayne Anderson, previous Enterprise Security Architect at McAfee.
Recently the Straight Talk Insights team at HCL Technologies invited a social panel to discuss a critical question at the center of today’s digital transitions: How do companies target investments and change the culture to avoid being the next victim of a cyberattack?
Alongside some fantastic leaders and technology strategists from HCL, Oracle, Clarify360, Duo Security, and TCDI, we explored the challenges of today’s hyper-connected and stretched security team.
Today, businesses operate in a world where over the last few years, more than 85% of business leaders surveyed by Dell and Dimensional Research say they believe security teams can better enable digital transformation initiatives if they are included early. Moreover, 90% say they can better enable the business if given more resources. Yet most of these same leaders assert that security is being brought in too late to enable digital transformation initiatives! These digital transformation trends—cloud, data, analytics, devices—are critical to the next generation of customer and employee experiences, and for the clear majority of companies, the transition of value chains is already in progress!
We collate the insights from the course of the discussion …
Q1: What are some of the IT security trends for 2019? Are there particular cybersecurity challenges related to digital trends?
Digital isn’t one trend—it’s many. Plus, we can’t stop running the business today. This forces a split of the skill investment that is available to companies, which MSSPs and system integrators can cover part of. The biggest challenge is information security extension in a multi-cloud world. All large enterprise is multi-cloud and hybrid. Yet few security operations teams are prepared for that.
Part of solving that challenge is bringing nascent ways of identifying anomalies and gaining scale—for example, through graph theory technology, critical to find the little traces that represent defensive capability. Machine learning will be throughout the information security technology stack soon. This shift must happen, as the challenge is more than new environments. The log volumes in cloud are material—and you pay for them, by the way—the formats are different, the collections are different, and the visibility is fragmented.
The harder thing here is that information security teams must adjust to ALL of this at ONCE. Great, you have AWS Cloud Trail. Let me ask you a question: Which of your security stack can see that AND is tuned for it AND can unify the risk identified there with on-premise derived visibility? And if you can answer that in a positive way, what about when I ask the same thing for Azure? Are you starting to think about the shift to resilience, or are you still thinking about defense and control exclusively?
I’d ask though, as your team is investing in cloud, are they investing in the understanding and readiness to protect data science? Are you preparing the project cycle for your security team to now be iterative as well to even deliver these services? Identity and access management is part of the solution as a critical foundation. Effective governance and strategy can help you figure out which platforms have security relevant data. While it’s easy to say “see and save everything,” you quickly find out how expensive that is, and how much trash is in there. At that point, you can start thinking about automation.
Focusing on data storage and data in motion has led us to consider more zero-trust to cut down on the amount of interstitial security complexity. To realize that vision, tokenization and indexing and many other technologies must continue to expand. We face an odd duality between the confidentiality and accessibility of making data useful in digital employee experience and customer experience.
It’s about more than adding automation to conquer the complexity. The automation must have intelligence, and it must operate in a way that is more than “I bought tech with buzzwords.” So many platforms and products say they do these things—but as you buy and implement, you need to focus on how, and how hard they are to build and link together. Plus, how are you going to maintain them? Be careful as we adjust to keep the pace of digital transformation that we aren’t trading one problem for another.
Finally, I’d note that at every level of the information security organization—not jus the CISO—the people need to have a sense of purpose. What value do you add as a security professional to the customer experience? Why do you exist? We need to remember that, as customer journeys are the way that digital transformation shows up. We have to think end-to-end.
Q2: What can companies do to protect themselves against vulnerabilities created by IoT devices?
Start with procurement. Look, I’d love to tell you that IoT security is a software problem, but that’s only part of it. It really starts with buying technology that is well-designed, and both the customer and the upstream vendor must enforce Secure Development Life Cycle (SDLC) internally.
To a certain degree, we need to see IoT as completely untrusted. Google’s BeyondCorp is a good goal for an entire org’s high-level vision of zero trust. Data introspection and device behaviors then need to have high inspection rather than assumptions of performance. We are advantaged in that we now live in a society full of tools where the reality is that encryption overhead is almost negligible with RISC based enhancements to network interface level assets. The organization can think differently about data protection in that kind of world with (relatively) cheap encryption cost to latency and performance.
When I think about IoT security, I continue to go back to an example that really made an impression on me a couple years back: If the team at IKEA can sell an IoT lightbar for cheap that has basic randomization, locked services, and minimal platform build … I have to think that certainly we can do better in health technology, industrial control systems, and manufacturing technologies.
When it comes to governance, IoT has the potential to turn asset management issues up to “11” on the 10-point scale of concern. How do you define an authorized device? Authorize an untrusted device to send data into the system? What do you recognize as a managed device? How will your organization make conditional access decisions to use, aggregate, and modify data? “Enterprise Architecture” (EA) needs to be part of the plan for effective governance. In some ways, as an industry, EA got swept up with the boom and bust of specific analyst models of architecture not proving out value cases at a lot of organizations. In today’s iterative digital world, architecture and simplicity have to be part of the IoT project Minimum Viable Product in order to realize the scale needed later.
We can’t manage IoT like laptops—these devices have fewer capabilities. Instead we need more affirmative approaches that integrate the components of the ecosystem in a predictable and defined way, like trusted cloud. The default expectation for a device intended to be used in a reduced management environment should have heavy encryption, PKI validation, and locked down application-controlled execution built into them out of the box.
When you take a step back and look at the problem as societal instead of the microcosm of a specific company’s product or implementation, public policy must enter into the intersection of law and devices at scale. We have to solve difficult questions like the role of liability and commercial incentives to build and deploy device platforms in a responsible way. As one example, when machine learning-led IoT decisions create a catastrophe, who is responsible? The owning company? The software vendor? The system integrator? All the above? In critical spaces like utilities and healthcare, we need the focus of meeting some level of criteria for devices to have minimum reasonable security.
Even at this scale, this, too could be a great place for graph theory and machine learning-led approaches to secure societal level device challenges like elections. It’s easily expressed as math—easily identified for loci and baseline deviations. We need investment, however, from government or non-traditional sources as the state/local government and education sectors have very long buying cycles, and the available budget for this problem hasn’t yet justified the extended R&D costs of these kinds of technological changes.
Even while these public policy shifts are emerging, the greater propensity of localized privacy law has created operational hurdles for enterprise. As a microcosm, introduction of privacy safeguards in the India data localization law represents many different interests trying to be balanced in one approach. This has created a higher cost for external multinationals as they create duplicative storage and has even slowed digital transformation and created a drag on growth for India based consulting and business process outsourcing economic engines. You could make the same analysis for CCPA or GDPR, but these same measures have helped privacy, potentially, for citizens.
To help companies navigate these challenges, we are seeing organizations like ENISA, and the NCSC Secure Authority providing advisory guidance. This leads to the definition of a state of reasonable practice. When we add that kind of practical dimension to ISO standards like the 27000 series, and the Top 20 from the Center for Internet Security, and others, we help organizations navigate what the basics look like for practical security applicability in IoT and security generally.
In Part II of this series, we’ll explore the threat of cryptocrime, the nature of cybersecurity threats in the near future, and the steps that small- and medium-sized businesses can take to protect themselves.
About the Author
Categories: Cloud Security