The Secure Operating System Equation
Many experts like the idea of a purpose-built, secure operating system. It's just that adopting one is not so straightforward, even if it's specifically for security-strapped SCADA systems
October 17, 2012
Hardened, secure operating systems for sensitive computing environments are nothing new. Trustix, SELinux, Sidewinder SecureOS, and Green Hills Integrity are among many secure OSes, some that have survived for niche environments and others that have faded into obscurity.
In 2008 Green Hills Software announced with much fanfare that its new Integrity-178B OS was certified as EAL6+ and was being sold commercially: EAL6+ is the highest security rating by the National Security Agency's certification program, denoting that the OS was designed and certified to defend against well-funded and sophisticated attackers.
And now the concept of secure OS is back in the limelight, with Kaspersky Lab founder and CEO Eugene Kaspersky yesterday confirming reports that his security company is working on creating a secure operating system specifically for SCADA and industrial-control system environments. In a post on his own blog yesterday, Kaspersky said the ideal way to secure these critical systems would entail rewriting all ICS software and incorporating the latest and greatest security technology. But that would be a massive, costly, and unlikely undertaking that still wouldn't necessarily lock down these systems. Hence the secure OS, he said.
"But there is a fully realizable alternative: a secure operating system, one onto which ICS can be installed, and which could be built into the existing infrastructure -- controlling 'healthy' existing systems and guaranteeing the receipt of reliable data reports on the systems' operation," he wrote.
The OS would be a purpose-built platform aimed at "solving a specific narrow task, and not intended for playing Half-Life on, editing your vacation videos, or blathering on social media," Kaspersky said. His company also is working on preventing third-party code execution or hacking of the OS. "This is the important bit: the impossibility of executing third-party code, or of breaking into the system or running unauthorized applications on our OS; and this is both provable and testable," he said.
Neither Kaspersky nor other company officials would reveal technical details, but a company blog post said that the OS would be written from the ground up and not based on existing code. Its kernel will be bare-bones when it comes to code, and clear of bugs: "...the core must be 100% verified as not permitting vulnerabilities or dual-purpose code. For the same reason, the kernel needs to contain a very bare minimum of code, and that means that the maximum possible quantity of code, including drivers, needs to be controlled by the core and be executed with low-level access rights," according to a company blog post.
Most security experts say they commend the concept of a secure OS for SCADA/ICS and other sensitive environments. SCADA software is notoriously vulnerability-ridden and dated, since much of it was built at a time when plants had no Internet exposure.
But the concept of a new secure OS is fraught with challenges when it comes to real-world implementation. "I like the idea that you have a firmware model where it's custom-built just for those [SCADA] devices," says Ken Silva, senior vice president for cyberstrategy at information technology contractor ManTech International. "The reality is that there are number of challenges associated with it: Where do you find developers for a purpose-built OS? Where do you find applications or monitoring solutions? There are all sorts of things that come with supporting an OS."
The advantage of a secure OS is that it runs fewer applications and tasks, so its attack surface is smaller than a conventional OS. "It's definitely more secure, but it's absolutely not foolproof," Silva says.
HD Moore, CSO of Rapid7 and chief architect of the Metasploit Project, says he wonders whether Kaspersky Lab may be building more of a custom security platform that hardens Windows rather than an all-new OS. "My gut feel is they don't mean OS. They mean a super-AV that does endpoint protection and HIPS stuff, and bundle that into a Windows 2008/2012 build and implement policies on top to limit the damage a bad application can do," Moore says. "That's my guess."
Even so, such a platform would still face some of the same challenges that a purely new OS would, he says.
SCADA vendor Siemens, meanwhile, says creating a secure OS for the SCADA world would require the open-source community to shore up their OSes, as well. "SCADA systems are highly tailored, developed for solving a specific narrow task. Siemens is working on methods of writing software, which, by design, won't be able to carry out any behind-the-scenes, undeclared activity," says Alan Cone, HMI product marketing manager at Siemens Industry Inc. "This is the important bit: the impossibility of executing third-party code, or of breaking into the system or running unauthorized applications on an operating system."
Cone says there's no silver bullet for fixing ICS security issues: It requires a defense-in-depth strategy. Some of the security layers Siemens provides today include user administration, secure communication, security devices, MAC filtering and blocking on switches, and IP block protection within the PLC code, he says.
Next Page: Secure OS 'worth a try' SCADA expert Dale Peterson, CEO of DigitalBond, says a new secure OS is worth a try. The trouble is that it won't change the inherent lack of security of SCADA systems: It won't add authentication to DCS or SCADA protocols, he says, or deter an application developer from hard-coding a backdoor account into software.
A few ICS vendors are using Windows 2008 Server Core, which has helped them shrink their attack surface, Peterson notes.
Kaspersky Lab, meanwhile, hinted that it may already be reaching out to the ICS vendor community as part of its development process. "It's a sophisticated project, and almost impracticable without active interaction with ICS operators and vendors. We can't reveal many details of the project now because of the confidentiality of such cooperation. And we don’t want to talk about some stuff so competitors won't jump on our ideas and nick the know-how. And then there are some details that will remain for certain customers’ eyes only forever, to ward off cyber-terrorist abuses," Eugene Kaspersky said in his blog.
[ Siemens has quietly made several security moves in the wake of Stuxnet's discovery two years ago -- most recently, new industrial control products that come with built-in security features. See Siemens Enhances Security In Post-Stuxnet SCADA World . ]
Secure OSes traditionally have not caught on in a big way, so Kaspersky Lab faces some big hurdles for their secure OS to fly, experts say.
"The idea of a secure operating system for SCADA is a noble goal, but market realities have prevented specialty secure operating systems from having much impact in the past," Rapid7's Moore says. "Customers simply don't care about the internal workings of the devices they purchase, and no secure operating system will make up for a poorly coded administrative interfaces. To date, many of the known SCADA vulnerabilities were caused by support backdoors, weak protocols, and generally bad design decisions by the vendor."
And security features could present overhead that could affect performance, too, he says. "Secure operating systems often depend on specific hardware features to implement things like nonexecutable pages and random number generators. Much of the hardware used by SCADA vendors is designed to be rugged, but not particularly advanced or fast. The overhead of security features may limit where this OS could be used."
Have a comment on this story? Please click "Add Your Comment" below. If you'd like to contact Dark Reading's editors directly, send us a message.
About the Author
You May Also Like