Is this the year when software-defined anything (SDx) becomes the template for federal agency IT security? It’s been knocking at the door for a while, and the spending outlook for government IT in President Barack Obama’s recent budget proposals could finally be the opening it needs.
In calling for a 35 percent increase in cybersecurity spending to $19 billion, the White House also proposed a $3.1 billion revolving fund to upgrade legacy IT throughout the government. Venting his frustration, and no doubt that of many others in the administration and Congress, Obama talked about ancient Cobol software running Social Security systems, archaic IRS systems and other old, broken machines and software at federal agencies.
That’s not a new story. Agency IT managers will readily tell you about the problems they have with trying to maintain legacy technology and the way that sucks up funds and manpower. They say they have too little time to focus on what they feel their jobs are really about, which is delivering better services to their users.
Security is just one item among many they must address, but it’s become a much more urgent one after a 2015 that saw major breaches at the Office of Personnel Management and elsewhere. That point was driven home again this year when the IRS revealed that over 100,000 attempts using stolen Social Security numbers had succeeded in generating the personal identification numbers used by tax payers to electronically file and pay taxes.
The revolving IT Modernization Fund in the White House budget proposal would pay for projects that will be prioritized based on the extent to which they lower the overall security risk of federal IT systems. The savings achieved by shifting to more cost-effective and scalable platforms will be recycled back into the fund.
Cost-effectiveness and scalablity are among the main advantages that proponents put forward for SDx architectures, along with agility in response to security threats. As threats become more targeted, more sophisticated and more numerous, protecting networks gets more difficult. With IT staff overwhelmed by just the legacy systems they have to keep running, organizations face much greater risk of damage from those attacks.
By simplifying infrastructure management with the software overlay that software-defined networking (SDN) brings, IT and security managers get a much better way of identifying when they are being attacked and a faster and more focused way of responding.
In a poll conducted earlier last year, ESG Research identified a significant percentage of enterprise security professionals who said they would use SDN to address network security across a wide range of different scenarios.
Researchers at the Idaho National Lab have already developed a proof-of-concept that uses SDx to emulate the use and security of the laboratory’s business systems. It’s already delivered “amazing outcomes” and demonstrates how SDx can be used to improve security, repeatability of processes and consistency in results, they said.
The future will only bring more security challenges for government, as the Internet of Things takes hold. That will introduce thousands of new avenues that attackers will use to try and penetrate networks. Given the kind of benefits that the IoT is expected to bring to government organizations, the trick will be in securing networks without limiting the facility of IoT.
One approach that won’t work is simply throwing the solution du jour at the problem, which has been the traditional answer. Bolting on more point-to-point, single-purpose devices simply won’t scale fast enough to deal with vulnerabilties and will be too costly. Those devices are also themselves proving more vulnerable than people thought, with Cisco joining Juniper and Fortinet in the list of manufacturers whose advanced firewalls apparently suffer from potential software problems.
Right now, the only viable solution in this brave new world of security seems to be through some kind of software-defined approach. It’s not a silver bullet by any means, and it must be part of an overall approach to security. IT and security professionals must also be convinced that it will provide for the kind of subtleties and granularity needed to weed out modern threats.
If -- and in an election year, it’s a big if -- Obama’s budget proposals make headway in Congress, SDx could prove the best way to tackle the security problems that otherwise threaten to overwhelm government.
Posted by Brian Robinson on Feb 12, 2016 at 10:46 AM0 comments
Government gets it in the neck frequently when it comes to cybersecurity, usually along the lines of it being too dense or too slow to react when problems arise. Some of that criticism is warranted, some not, but let’s give credit where credit is due.
House lawmakers were quick to jump on the revelation that Juniper Networks, which sells its popular NetScreen firewalls to many government agencies, had found flaws in the operating system that runs those firewalls. This defect would allow someone to remotely access a device through SSH protocols or telnet and then monitor and decrypt VPN traffic.
On Jan. 21, the House Committee on Oversight and Government Reform sent letters to the heads of major agencies asking them to audit their use of Juniper’s firewalls and report by Feb. 4 on how they might have been affected by the ScreenOS flaws and what corrective measures they took prior to Juniper releasing a software patch on Dec. 20.
The committee’s fast action follows the devastating breach at the Office of Personnel Management last year, which went undetected for several months. A year earlier, major problems were found with the widely used OpenSSL protocol, which may still be affecting systems around the world today.
It will be interesting to see what the House committee finds. Any agency that is on top of its security game should already have done that Juniper audit and should have no problem providing the information requested. Those that haven’t may have to scramble, and any committee report should show the extent of that.
Other elements of the government’s security status aren’t developing so quickly. Last year, the Government Accountability Office gave its regular report on the status of government cybersecurity, giving a lukewarm review of the Department of Homeland Security’s EINSTEIN program, more formerly known as the National Cybersecurity Protection System (NCPS).
EINSTEIN was designed some years ago to be a central plank in the government’s overall cybersecurity posture, aimed at providing agencies with intrusion detection, intrusion prevention, analytics and information sharing technologies. If those tools were fully in place across agencies, breaches such as those at the OPM and other agencies may have been prevented, or at least noticed and mitigated much sooner than they were.
Getting EINSTEIN in place governmentwide has been frustratingly slow, however, and according to the latest GAO report on the system, that sluggish pace continues. The DHS program is still only partially meeting its objectives, GAO said, and is deficient in all four areas examined.
With intrusion detection, for example, it can only compare network traffic to known signatures of malware, which covers maybe 80 percent of the bad stuff. The other malicious activity, which contains the advanced persistent threats that do most of the damage these days, requires more sophisticated detection.
Likewise, EINSTEIN now only prevents intrusion of particular kinds of malicious data, but it can’t block the kind that’s hidden inside the web traffic itself. DHS says it plans to deliver that capability sometime this year.
Overall, the uptake of EINSTEIN has been spotty, because of deficiencies at the agencies or the DHS itself. All of the 23 agencies required to implement intrusion detection capabilities had routed at least some of their traffic through the NCPS sensors, the GAO said, but only five were receiving intrusion prevention services. Agencies had not taken all of the technical steps needed to implement the system, in part because the DHS had not yet provided them with the necessary guidance.
It’s all an example of the strange and often puzzling disparities in the government’s approach to security. On the one hand, at least some parts of Congress seem to understand the urgency and are prepared to pressure agencies to move faster. On the other, critical technology that was recognized as essential years ago still isn’t fully deployed.
Posted by Brian Robinson on Jan 29, 2016 at 12:01 PM0 comments
Is the Secure Shell (SSH) vulnerability going to be this year’s OpenSSL? As with the stock market, it’s a mug’s game to predict the future, but warning flags have been raised in response to reports of problems with major security devices.
It was problems with the OpenSSL version of the Secure Sockets Layer encryption that led to the discovery two years ago of the Heartbleed bug, which many security professionals called one of the scariest things they had seen. It allowed anyone who could get to an infected device to compromise the private keys used to identify service providers and encrypt data traffic.
Eventually, hundreds of thousands of servers around the world were found to be vulnerable to Heartbleed, and even now no one seems sure if all the holes have been plugged.
In December 2015, Juniper Networks said it had found “unauthorized code” in its ScreenOS, the operating system that runs on its widely used NetScreen firewalls. That code would allow a knowledgeable attacker to gain administrative access to NetScreen devices over SSH and Telnet, the company said, and to decrypt VPN connections.
The company has since made several fixes to its software to close down the gap, the latest to the Dual_EC random number generator used in the firewalls. That’s been a long time coming, since Dual_EC has reportedly contained a backdoor inspired by the National Security Agency (that could also be exploited by bad guys).
Now researchers have found suspicious code in Fortinet’s FortiOS firewalls, saying it was also essentially an SSH backdoor. Fortinet, however, downplayed that allegation, saying it was a “management authentication issue” that had been fixed some time ago.
Coincidentally, the National Institute of Standards and Technology recently released a new guidance document on the security of SSH key-based access, which it said is often overlooked by organizations. That would be a bad thing, as NIST also points out, because misuse of SSH keys “could lead to unauthorized access, often with high privileges.” In other words, it’s potentially handing the keys to the kingdom over to people who will gratefully accept the gift -- and then take you for all you are worth.
Backdoor keys are specifically mentioned by NIST as one of the seven categories of vulnerability in SSH, which is widely used to manage servers, routers and other security devices as well as firewalls. It’s also used to provide privileged access to servers and networks.
However, NIST pointed out, SSH public key authentication can also be used to create a backdoor by generating a new key pair and adding a new authorized key to an authorized keys file. That allows someone to get around the access management system and its monitoring and auditing capabilities.
Other vulnerabilities NIST cited include: poor SSH implementation; improperly configured access controls; stolen, leaked, derived and unterminated keys; unintended usage of keys; theft of keys as attackers inside the system move from server to server and steal credentials along the way; and the always present human error.
The recent firewall revelations are by no means the only reported problems with Secure Shell. In the middle of last year, researchers also discovered vulnerabilities with the OpenSSH version of the protocol, which allowed attackers to get around authentication attempt limits and launch brute force attacks on targeted servers.
The big problem with these kinds of vulnerabilities is not necessarily that they exist. If they are quickly noticed and patched, any likely damage is minimized. But the OpenSSL bug went unnoticed for several years, so the door to networks and systems that used that protocol was open all that time. The OpenSSH bug could have been present on versions of the FreeBSD operating system as far back as 2007.
Heartbleed redux? Not so far, it seems, but the year is yet young.
Posted by Brian Robinson on Jan 19, 2016 at 1:56 PM0 comments
Looking back, 2015 was a time of strain in the public sector when it came to cybersecurity, with the hack of systems at the Office of Personnel Management that exposed over 20 million government employee records, the high infection rate of state and local networks by malware and ransomware and the overall lack of security compliance of government software. Here, then, is a start-of-the-year list of some the issues government will face in 2016.
The Internet of Things: The rap about the IoT is that it’s still on the horizon, but many argue that it’s already here, and people are putting (betting?) money on it influencing both industrial IT and the evolution of smart cities. The White House demonstrated its support when it jumped into the fray in 2015 with a $160 million initiative. But there are concerns that the technical evolution of the IoT is getting too far ahead of the much slower development of security and privacy policies. This year should see rapid movement around those issues. We hope.
Contractors get the security eye: Big integrators and government contractors have long been aware of the need for tight cybersecurity in their work with government agencies, but the same can’t always be said of the subcontractors they hire. As hackers get more sophisticated about how to access government systems, they are finding vulnerabilities in the security of those smaller firms, with devastating results. The OPM hack, for example, was attributed to the theft of a vendor’s network access credentials. The massive breach at the retailing giant Target the previous year was blamed on credentials stole from a HVAC vendor. The government has begun to try and rein in lax vendor security, with the Office of Management and Budget issuing “cyber guidance” for contractors. Don’t bet on that being the last word, however.
Encryption: There’s a debate brewing over encryption. On the one hand, it’s considered essential to cover security, given that best practices now assume that persistent hackers will penetrate even the best defenses. In that case, the more data and communications can be encrypted, the safer they will be. OMB seemed to be following that logic when it issued a memo in June requiring federal agencies to encrypt their websites and web services connections. But others say that it’s a much more nuanced argument than just encrypting everything, and that true security also requires a way to inspect both incoming and outgoing encrypted traffic. Email encryption has become a particular target for security critics, especially in light of the hackers of Pentagon email networks who reportedly took advantage of outgoing encrypted traffic that was not being inspected. Meanwhile, the government is having to address public demands for even more encryption, a move intelligence agencies view with suspicion, saying it could hurt their efforts against terrorism.
The value of NIST: The National Institute of Standards and Technology has never been an outfit to blow its own horn, so its impact on cybersecurity over the past few years has often seemed less obvious than that of the Department of Homeland Security and DOD. But NIST is arguably the most ambitious agency when it comes to addressing the more technical aspects of cybersecurity, and 2016 could finally be the year when agency gets the recognition it deserves. Surveys already suggest that the majority of government agencies are now getting on board with NIST’s Cybersecurity Framework, and the private sector is also paying greater attention to that guidance. Other NIST endeavors, such as its Identity Ecosystem Framework, the first version of which was released in October, are also taking flight. The agency recently asked for proposals that can develop identity solutions for state and local government.
The malware ecosystem: The biggest hurdle government may have when it comes to cybersecurity is realizing what it’s up against. Despite the now-clichéd idea that only state actors such as Russia and China are capable of the biggest and most penetrating attacks, the lure of filthy lucre has created a widespread and highly networked ecosystem of criminals and hackers that crank out highly sophisticated threats. And government is not ready to deal with this kind of industrialized threat machine, according to some observers. The spectrum of tools now available to hackers is immense and growing, and the research and development that goes on in this underground industry is impressive -- turning even legitimate proxy networks into channels hackers can use to their advantage.
The year security is taken seriously? Despite 2015 providing an embarrassment of cybersecurity breach riches, there’s no sign that government overall is paying that much more attention to information security -- it still seems to fall behind many other issues that cash-strapped agencies must address. Whatever security resources are being deployed seem to be mostly aimed at reaction rather than preemption. However, there are signs that attitude may be changing. Early in 2015, the DOD said it would kick workers guilty of “poor cyber hygiene” off networks they need to do their jobs, and Congress has made noises about stricter oversight of agency cybersecurity and holding agencies accountable for failures. Time will tell if these efforts amount to anything. If the lesson of the OPM hack -- that agency executives can lose their jobs if they don’t take care of security -- doesn’t hit home, you have to wonder what will.
Posted by Brian Robinson on Jan 04, 2016 at 12:48 PM0 comments