Security threats to virtualized environments is not a new subject, but it’s one that should be gaining prominence as organizations, and particularly those in government, virtualize more in order to cut costs and improve IT efficiencies. If agencies don’t consider the security implications, they’re opening themselves up to a world of hurt.
That’s especially true since the world of malware innovators is not standing still, as a new report from Symantec points out. As fast as defenses are erected, attackers come up with ways to get around them.
One of the more recent exploits involves attacks that are designed to wait out the automatic malware detection and analysis defenses that are increasingly being built into virtual systems. Some trojans will simply wait for multiple mouse clicks to occur before they decrypt themselves and start up their payload, and that can make it all but impossible for automated systems to come to any timely conclusion about the threat.
“Time is on [the malware’s] side,” said Candid Wüest, author of the Symantec report. “If the sample does not behave maliciously within the first five to 10 minutes, the analysis system will most likely deem the file as harmless.”
This in turn has prompted attackers to develop other methods of evading automated analysis on virtual machines, such as focusing more on the user’s interaction. The malware waits, for example, for three left-button clicks on the mouse before executing. In that case any kind of user interaction, say the use of a CAPTCHA box, a test to determine if a user is human, could prompt action.
Those kinds of exploits are harder to patch on virtual machines and require some background monitoring to generate the necessary interaction triggers, Wüest said.
The most complete government guide to virtualization security is NIST’s three-year old SP 800-125 Guide to Security for Full Virtualization Technologies. It says that the security of a full virtualization solution is heavily dependent on the individual security of its components – from hypervisor and host operation systems to applications and storage. Other sound security practices, such as keeping up to date with security patches, are also necessary.
That’s all true, but it doesn’t seem to be enough in the face of what appears to be an inevitable push by malware designers into the virtual space. And it could be that virtualization attacks are now an embedded feature in most malware. Up to 82 percent of the malware tracked by Symantec was able to run on virtual machines.
It’s not as if organizations haven’t been warned. As long ago as 2009, the Cloudburst Attack showed how attackers could go through a guest virtual machine to attack the host, in many ways an IT administrator’s worst nightmare. In 2012, the Crisis malware, which targeted Windows and Mac systems, was shown to also be capable of sneaking onto virtual machines if a specific image was installed on it.
Given the range of threats now facing virtualized environments, Symantec recommends a number of best practices for organizations to follow:
- Protect the host server, which provides access to virtual machines, with lockdown solutions and host intrusion detection systems along with regular software updates and patches.
- Protect both the host server and virtual machines running on it with “proactive” components that go beyond classic security such as antivirus scanners.
- Have administrators enforce proper user access controls to the servers hosting virtual machines and use two-factor authentication or other strong login processes.
- Make sure the virtual machines are fully integrated into disaster recovery and business continuity plans.
- Network security tools should also have access to the virtual network traffic between the virtual machines.
- Snapshots and images of virtual machines need to be included in the patch and upgrade cycle. Unpatched virtual machines are frequent targets of malware.
- Integrate virtual machines into the security logging and security information and event management (SIEM) systems that are used for all other IT devices.
The overall message, which NIST stresses, is that the security of virtual machines and networks needs to be handled just as intensely as that of other IT. Given the rate of innovation on the malware side, that might not be all that’s needed, but it will go a long way.
Posted by Brian Robinson on Aug 29, 2014 at 11:58 AM0 comments
This month marks the 10th anniversary of Homeland Security Presidential Directive 12 mandating the development and use of an interoperable smart ID card for civilian government employees and contractors. The results of the program so far range from the impressive to the disappointing.
“I would call the programmatic platform a huge success,” said Ken Ammon, chief strategy officer for Xceedium, an ID management software vendor.
As of the first quarter of this year, 5.4 million Personal Identity Verification (PIV) cards have been issued to civilian employees and contractors, accounting for 96 percent of those who need the cards. Given employee turnover and the need to periodically reissue the cards, the coverage is quite good.
The challenge now is having them used as they were intended, as strong, two-factor authentication for both logical and physical access across agencies. This is a multifaceted challenge that is proving to be a much tougher nut to crack than designing and issuing the cards.
HSPD-12 was issued Aug. 27, 2004, by then-President George W. Bush. The heart of the mandate was simple. Inconsistencies in government IDs left the government vulnerable to terrorist attack. “Therefore, it is the policy of the United States to enhance security, increase government efficiency, reduce identity fraud and protect personal privacy by establishing a mandatory, governmentwide standard for secure and reliable forms of identification issued by the federal government to its employees and contractors (including contractor employees).”
The National Institute of Standards and Technology was given six months to produce the standards, which included identity vetting and secure, interoperable digital technology. Eight months after that, agencies would have to require use of the cards, “to the maximum extent practicable,” for access both to physical facilities and IT systems.
The first part of this effort, developing the standard and technical specifications and designing, producing and issuing the PIV cards, is the programmatic success Ammon cited. But the second part, the qualification “to the maximum extent practicable,” has proved to be a speed bump.
Seven years after the directive, the Government Accountability Office concluded in 2011 that although substantial progress had been made in issuing PIV cards and fair progress in using them for physical access to government facilities, only limited progress had been made in using them for access to government networks and minimal progress in cross-agency acceptance.
A year later, increasing the use of PIV and the military’s Common Access Card credentials was identified by the White House as a priority area for improvement. Agencies were given until March 31, 2012, to develop policies for the use of these credentials.
Reasons for the lack of widespread use cited by GAO were not technical, but administrative: Logistics, agency priorities and of course budgets. “According to agency officials, a lack of funding has . . . slowed the use of PIV credentials,” the report stated.
But technology also is an issue, as the card is only one element in any authentication system. Use of the electronic credentials in the cards has to be incorporated into systems already in place, or those systems must be replaced. Under a 2011 White House directive, all new systems under development at agencies must be enabled for PIV credentials and existing systems were to be upgraded by fiscal 2012.
Like many unfunded mandates, this has been a tough one to meet. And in the meantime, technology keeps changing. Mobile computing, for instance, means that many government workers are using tablets and cell phones for work. Technically these should require PIV authentication for government work, but many are not equipped to accommodate that.
HSPD-12 is not a failure, but it could be doing a lot better if strong, two-factor authentication was a higher priority within agencies. However, the the rapid pace of technological change makes it unlikely that any government-mandated technology will ever be completely successful. Even so, much more could be done.
Posted by William Jackson on Aug 22, 2014 at 10:16 AM2 comments
On a global basis, Android devices far outsell those that use other operating systems. But it’s been a much different story in government, where Apple has become a preferred mobile device supplier in many cases and where Blackberry still has a strong presence.
The situation is caused mainly by perceptions that Android security is suspect. But that may finally be changing, based on work by Samsung, the leading smartphone supplier. Its Knox containerization technology, under development for four years, seems to be gaining traction across the federal, state and local markets.
Now government pilot projects are being launched and, according to Samsung, attracting potential users who are coming to see how they can use the technology.
“We have numerous examples of where agencies are willing to enter into those initial presentation pilots,” said Johnny Overcast, director of government sales for Samsung Mobile. “We’ve been working with major executive branch agencies in particular for some time, and there have already been significant purchases of Samsung Knox.”
The Defense Department was one of the first to get on board with Android, and Samsung in particular. In May 2013, the Defense Information Systems Agency announced Security Technical Implementation Guides (STIGs) for mobile devices aimed at getting the technology into the hands of military users as quickly as possible. The STIGs describe the security policy and configuration requirements for government-issued devices, including those that use Samsung Knox.
More recently, the Army announced it would use Samsung Galaxy Note II smartphones as the end user device in its Nett Warrior program, whose goal is to give front-line soldiers advanced situational awareness capabilities.
In March, the DOD approved the Samsung Knox Hypervisor virtualization technology and Authority to Operate on sensitive networks.
The departments of Justice and Homeland Security have also bought into the Knox hardened approach for Android, along with various three-letter intelligence agencies.
To some extent, Samsung Knox closes a circle, since it uses the Security Enhanced (SE) Android specification developed by the National Security Agency, which prevents any user without proper permission from getting access to the secure container. It also extended the use of the NSA’s SELinux into the Android operating system. An even further closing will happen when Google integrates elements of Samsung Knox into Android L, a next-generation version of the operating system that had its beta release in June.
Samsung Knox adds to the secure capabilities that Android already has, Overcast pointed out. Vanilla Android, an install of Android without customization, already offers discretionary user access control, and the Knox platform adds such things as a trusted boot process.
That trusted boot uses Trust Zone Integrity Management Architecture to continually scan the hardware, applying a mathematical check to make sure that what’s being loaded onto the device is authorized.
With the technical basis for Samsung Knox increasingly accepted by users, Overcast said the company is focusing on broadening the choices those users will have. It now provides for multiple user domains on a device, for example, and the ability for users to choose what kind of container technology they have. With Knox’s new multiuser framework, administrators can also select what permissions and applications can be used with specific containers.
This is all in preparation for what Overcast said he sees as a tipping point in the government mobile markets, when agencies get beyond fundamental questions about security and instead look to the kinds of devices that will help them best execute their missions and provide better services to citizens.
And that, he said, is not too far into the future.
Posted by Brian Robinson on Aug 15, 2014 at 11:07 AM2 comments
The General Services Administration has spelled out a new policy for agency IT projects to ensure that basic principles promoting economy, efficiency and transparency are integrated into technology solutions developed for or operated by GSA.
Included in the IT Integration policy issued July 24 are requirements that cybersecurity be incorporated into IT projects from the beginning and that the appropriate security team has a place at the table during planning.
“One of the largest challenges for GSA IT is early and consistent engagement with the IT security team throughout the project to understand what security requirements apply, who needs to be engaged to assist in implementation and how this impacts the project schedule,” agency CIO Sonny Hashmi wrote in the instruction letter.
With the cyber threat landscape growing in intensity and sophistication, security no longer can be layered on in IT projects as an afterthought, Hashmi explained in a blog post. “This principle will require that the GSA Office of the Chief Information Security Officer acts as a consultant and partner throughout the project life cycle, rather than being viewed as a compliance step towards the end of the project,” he wrote.
Hashmi also spelled out another principle that could help significantly improve cybersecurity: platform reuse first. That is, GSA will give priority to leveraging existing platforms for new services over building new systems.
Cybersecurity is just one part of the new GSA directive. It also includes compliance with the federal cloud-first policy and requirements for a GSA open-source-first policy as well as for single sign-on, online delivery of services, records management and better stewardship of procurements.
But I am focusing on the security requirements. IT security has been designated by the General Accountability Office as a high-risk area for all executive branch agencies since 1997 and has remained so since. This is not so much because there has been no improvement in security, but because government dependence on IT continues to increase as the systems become more complex, making it difficult for administrators to keep up with security requirements.
Ensuring that security is included from the earliest stages of planning and development could help change this. Reusing existing platforms to reduce the number of new projects being developed also could improve security by allowing administrators to concentrate on a smaller number of legacy systems with a known security profile. Expanding the use of existing platforms does not guarantee their security, of course. Expansion and repurposing will require new evaluations and new controls to make sure they meet risk management requirements. But if done well, this could be more efficient that constantly bringing new systems online.
The new policies apply to all new GSA projects, regardless of size, to all enhancements of existing systems that are over the $150,000 threshold for simplified acquisition and to any cloud acquisition or blanket purchase agreement regardless of value. Failure to follow policy could result in project termination.
Like any policy, GSA’s IT Integration policy could devolve into a morass of paperwork and checkboxes that achieves little or nothing. But if the new policy cuts through existing bureaucracy rather than add new layers, it could be a step toward improving the agency’s cybersecurity.
Posted by William Jackson on Aug 08, 2014 at 10:06 AM0 comments