Six burning questions about network security
- 06 June, 2008 09:56
Security issues often seem to smolder more than burn, but these six are certainly capable of lighting a fire under IT professionals at a moment's notice. Handle with care.
Is server virtualization worth the risk?
The benefits of moving away from traditional servers to virtual-machine (VM) arrangements are the cost savings in hardware consolidation and remarkable flexibility. But less-welcome consequences can be security gaps and virtual-server sprawl, risks that draw fire from auditors.
VM security too often is being addressed after the fact, says Douglas Drew, senior consultant with BT's emerging technologies office and an auditor for the Payment Card Industry (PCI) standard. "How do you handle access control or auditing? Suppose I migrate an instance of a virtual machine from rack A to rack B: Is one a locking rack that needs a physical badge to get to the console and the other not? Does the VM hypervisor allow for separation of administrators A and B so A can only logically touch systems A and administrator B only touches B? How are you re-upping the risk assessment based on the architecture change?"
Like more traditional networks, the VM environment, whether based on VMware, XenSource or Microsoft, calls for applying best practices defined under the ISO 27002 standard for secure systems, Drew says. "We've seen some cases where people are slow to adopt VM because they haven't gotten their arms around this."
And VM software out of the box won't suffice for security, many say.
"The virtual machines are mobile, they're designed to be mobile," says David Lynch, vice president of marketing at Embotics, a start-up that makes VM life-cycle management software. "You take a physical server and make a clone of it. You lose the identity of the physical server, but your existing management tools are based on the idea you have a physical server."
As designed today, VMware's VirtualCenter management won't prevent VM prawl because VM ID numbers can be changedand re-set, Lynch contends. He adds it's not possible to ensure a unique VM ID system for an enterprise using more than one VirtualCenter.
The Embotics software, which works with VirtualCenter, tries to compensate by using a cryptographic hash, combined with VM meta-data, to brand a VM ID as legitimate and authentic. Other start-ups, including Fortisphere and ManageIQ, also are tackling the VM sprawl issue.
Some security vendors are convinced that the main VM software developers are in such a rush to get their products out to grab market share that as Andrew Hay, product program manager at Q1 Labs, puts it, "security is an afterthought."
Hay notes there's no Netflow-enabled virtual switch to help with activity monitoring. "You're creating a separate network that happens to reside on a box," Hay says. "But no one pushes for flow analysis in the virtualized world."
Should all this stop IT managers from going virtual? The bottom line, according to Hay: "It would be best to research your options before going fill tilt."
Does stopping data leaks lure lawyers?
Data-loss prevention (DLP) -- or call it data-leak protection -- lets you monitor content for unauthorized transmission. But organizations gaining experience with it are finding DLP sheds so much light into the darker corners of the corporate network that IT and business managers may find themselves in regulatory and legal peril.
"You move from ignorance to compliance jeopardy," says Tony Spinelli, senior vice president of information security at credit information services firm Equifax, describing the early days of deploying the Symantec DLP in his organization. DLP became a spotlight in the dark, exposing data-storage practices that needed to be improved.
That puts both business and IT management on the spot to make changes. And more security managers are finding that picky auditors -- once they know the DLP tools are in place -- are demanding security changes that corporations would be at their legal peril to ignore.
So is this "see all, know all" aspect -- in addition to the fact that DLP can still be expensive -- enough of a downside to turn off potential buyers? Maybe, but it would mean turning away from the most promising content-monitoring approach, one that might be sorely needed to help keep your firm out of regulatory and legal trouble.
Knowing in advance that DLP may be a disruptive technology, security managers can make plans to prepare business managers -- the rightful data owners in the eyes of most corporations -- as well as auditors and legal staff.
One security professional with DLP experience, Ron Baklarz, who recently left MedStar Health to join Amtrak as chief information systems officer, said the approach he took with the Reconnex DLP used at MedStar was to bring business people into the data-oversight process.
"You need to partner with them on compliance," Baklarz advises. Giving authorized business staff a log-in to technical DLP systems makes them active participants in the data-loss prevention effort.
In-the-cloud security: Dreamy or dangerous?
According to John Pescatore, a Gartner security expert who keeps tabs on in-the-cloud security services, the basic thing about them -- be they be for e-mail, denial-of-service (DoS) protection, vulnerability scanning or Web filtering -- is that they're an alternative to the do-it-yourself approach in buying software or equipment.
There are strong reasons to ascend into the cloud by buying a service -- but also times to stay in the more earthly domain with your own stuff.
To start, it's worth thinking about enterprise in-the-cloud security services as two basic types, Pescatore suggests. The first is bandwidth-based, such as carrier- or ISP-based DoS protection and response.
"AT&T, for example, can do this better and more cheaply than you can, plus they're filtering out attacks further upstream than you can using their bandwidth," Pescatore says. The alternative would be buying anti-DoS equipment from a firm such as Arbor Networks and setting up protection on your own.
The second in-the-cloud type is what Gartner prefers to call"security as a service," which is "totally divorced from a bandwidth service," Pescatore says. Using an antispam service, for example, involves redirecting the MX record to the service provider but doesn't entail specific bandwidth services tied to one single carrier.
This genre includes e-mail spam and antivirus filtering; vulnerability scanning; and Web filtering. What it doesn't include, by and large, is either DLP content monitoring and filtering or identity access and management, which are tightly coupled to internal business changes.
Using security-as-a service in the cloud makes a lot of sense to protect mobile laptops or provide protection for widely distributed branch offices, Pescatore says."For the very large global corporations, this is attractive," he says.
However, most companies will probably find it as easy and cost-effective to continue to guard internal operations by deploying their own security gear to filter spam, viruses and restrict Web access.
There are potential risks to filtering services. You might not want to transmit sensitive business transactions through this kind of third-party service. And, there's always a chance the service might go out of business.
All of these in-the-cloud services are still fairly new, seeing a growth spurt only in the past three years, Pescatore says, with MessageLabs, Microsoft, Google's Postini and Websense to be counted among the vendors. Gartner estimates that in-the-cloud e-mail security services don't account for more than 20 per cent of the total e-mail security market but will jump to 35 per cent by year-end and 70 per cent by 2013.
According to research firm IDC, last year the market for e-mail security software was US$1.38 billion, and appliances another US$692.2 million. In-the-cloud services, which IDC calls hosted services, were US$454 million. Software and appliances are expected to continue steady growth (see chart), and hosted services will jump to US$638 million this year and US$1.39 billion by 2011.
This kind of expansion of in-the-cloud services encourages the Jericho Forum, an organization of about 60 corporations which has been actively pushing for innovative e-commerce security that reaches outside the traditional corporate boundary of the perimeter.
"Web filtering in the cloud has only taken off in the last 16 months," says Paul Simmonds, a member of the Jericho Forum board."There are many more in-the-cloud services today than there were a few years ago." Simmonds said the"disappearing perimeter" in corporate networks is making in-the-cloud security services an appealing option that many businesses are exploring today.
Will Microsoft ever get security right?
When it comes to security, can Microsoft get any respect?
Even Bill Gates has humbled himself at times to explain why Microsoft has fumbled the ball so often on security. In his last public appearance at the RSA 2007 Conference, on stage with Craig Mundie, to whom he handed the baton to direct product security going forward, the two offered a mea culpa explanation on why Microsoft's software has fallen short.
"Humans are humans and they make mistakes," said Mundie, with the duo later indicating that the inadequate security plaguing Microsoft software in the past can be traced to a naivete in the early years based on the perception few controls were needed because"everybody was really good" and the data center seemed carefully tucked away.
This decades-old baggage remains a burden for Microsoft, says Andreas Antonopoulos, senior vice president and founding partner at Nemertes Research.
"Even today, the fundamental design decisions made 25 years ago still haunt Microsoft," Antonopoulos says."Windows Vista is not a new operating system; there are a lot of the older operating systems under the cover, which carries with it the baggage of the last 20 years to ensure backward compatibility of applications."
Microsoft is caught in a conundrum, Antonopoulos asserts. If the company really decides to make a fresh start on software, it would likely have to sacrifice financial advantages."That's not likely to happen," he says.
Burton Group analyst Dan Blum expresses a similar opinion, saying,"They are compromised in that sense. They have the constraints of backward compatibility in mind."
A few years ago, Microsoft sought to make a break with the past in what was called the Next-Generation Secure Computing Base (NGSCB) project, but"they killed it," Blum says.
According to Blum, Microsoft remains driven along a path of"convenience and flexibility and backwards compatibility," which gives them the best advantage in the marketplace. Blum muses if he were Bill Gates, he would have tried a"parallel approach" to develop a next-generation trusted operating system, even if it broke with existing applications.
In other respects, Microsoft has produced a viable identity strategy overall, Blum says, but it's been hobbled by its Windows-centric approach."They never put a stamp on any platform that's different," he says. Microsoft's lack of support for the SAML standard, for instance,"is a big mistake and not in the best interests of the industry."
Linux, Unix and Macintosh operating systems ship with better"secure by design postures" than those from Microsoft, Antonopoulos contends.
But Antonopoulos and Blum both say Microsoft has improved with Vista and XP2.
"The problem is Microsoft has developed a bad reputation and it's hard to outlive that," Antonopoulos says. Microsoft has plenty of talented engineers with identity and trust expertise, but their ideas expressed during engineering conferences seldom seem to get adopted in Microsoft software."I think they must get overruled."
For some third-party security software providers that work closely with Microsoft, it's also been trying at times.
"It's been a roller coaster," says Phil Lieberman, president of Lieberman Software, which makes password and administrative management tools that work with Microsoft desktop and server products."The problem with Microsoft is it's not just one company; it's divergent ones on different paths fighting each other."
In some Microsoft units, such as those managing CRM or Office products, there's no effort to work with third-party applications for security while"the core operating system group is more open," Lieberman says.
But the most aggravating part of working with Microsoft -- which may be necessary to gain official Microsoft certification -- is that the company isn't keeping up on the technical documentation.
"A tremendous amount of the operating system is undocumented," Lieberman says."They're moving so fast and doing so many releases and updates, no one is keeping track of what they're doing. For instance, if Microsoft goes and changes something for Patch Tuesday, and a [Data Link Library] is changed, they don't bother to change the documentation, and your application stops working. We have to go research this and we find they've changed it."
While acknowledging Microsoft's poor track record, others are a tad more conciliatory.
Microsoft's efforts to improve have had a"positive impact," says Oliver Friedrichs, director of Symantec's Security Response division."We have to give Microsoft some credit for improving operating system security."
In the past few years there just haven't been the types of devastating worm attacks, such as Code Red, Blaster and Nimbda, that exploited holes in Microsoft products to wreck havoc around the world.
"Attackers today are focused on the third-party Web plug-ins," Friedrichs adds.
"It's easy to pick on Microsoft because they're ubiquitous and historically had a problem," says Jon Gossels of SystemExperts."But year after year, their products are getting better, and a lot of professionals out there are trying to find the bugs."
NAC: Is your firewall enough?
Network access control (NAC) isn't for everybody, but it can be a valuable tool for controlling the circumstances under which individuals gain network access.
That can be valuable for heavily regulated businesses. NAC can perform a comprehensive check of endpoints before they are allowed to get on to corporate networks, and that kind of check can help placate regulators that demand enforcement of policies about how legitimate endpoints must be configured.
Most NAC platforms not only perform this function, but they keep records of performing it, something demanded by various regulations such as payment card industry (PCI) standards and the Health Insurance Portability and Accountability Act (HIPAA).
Screening guest users is a particular problem that NAC can address well, according to Gartner."Most Gartner clients that are planning to deploy NAC report that their first priority is to implement a guest network," says Gartner analyst Lawrence Orans in a report."In 2007, many security managers who viewed NAC as a strategic security process were able to use the near-term benefits of guest networking to justify getting started in NAC."
If businesses have a diverse set of full-time employees, contractors and guests that use their networks regularly, NAC can help assure that the devices they use to connect meet configuration policies. For machines that flunk, NAC can either fix them, quarantine them or grant them access to a network segment with only limited resources and where they can do limited damage.
Similarly, businesses that need to segment their networks based on department or job function can use the authorization controls in NAC to do so on a fairly detailed level.
"We see a perfect storm if companies have multiple compliance requirements (the Sarbanes-Oxley Act, PCI, HIPAA), a diverse workforce (employees, contractors, remote workers, partners, suppliers) and global operations (the need to segment the environment by region, business unit and others)," says Rob Whitely, an analyst with Forrester Research.
NAC will ultimately become an element of layered security architectures that rely less on perimeter firewalls as the main bastion and more on layers of security that seek to mitigate threats, Whiteley says."This is part of a larger trend around de-perimterization. NAC is not necessary, but will become a critical component for these new security architectures," he says.
Most networks can get by without NAC. The technology reduces the risk that compromised machines gain network access and that they can do damage if they manage to get admitted anyway. But it doesn't guarantee security. NAC came about in response to threats that traditional Layer 3 firewalls couldn't handle, and there are threats that NAC can't handle, but it can make important contributions.
"Ask yourself: Is your firewall enough?" Whiteley says."If so, NAC is most likely unnecessary. It provides the additional host integrity checking, but this provides little value above and beyond more granular authentication and authorization -- which are really just attempts to make up for shortcomings of today's firewalls."
Has IT licked patch management?
Patch and vulnerability management tools can take on detecting and protecting vulnerable machines in a mostly static, controlled environment. The technology area is considered a priority among IT managers, who most likely have spent many years perfecting their vulnerability scanning, patch testing and software distribution processes.
According to an Enterprise Management Associates (EMA) survey, more than three-quarters (76 per cent) of 250 IT managers surveyed had some sort of patch management product in house and said patching is an important to very important process. Van Dyke Software's fifth annual enterprise security survey of 300 network administrators showed that 30 per cent still worry about patching, a number that has declined over the years. Some industry watchers speculate that the lessening concern reflects a maturity in patch management products and in IT managers' processes.
"We really don't have anything that changes what needs to be patched, with the exception of remote access users, who are constantly a difficulty for us to keep patched," says Craig Bush, network administrator at Exactech in Gainesville, Fla. "Currently, we have to wait until the client connects to our VPN to make the updates happen, which isn't always regular."
Bush says his patching processes are mature, but vendors could ease the process by building in time for users to properly test patches before rolling them out across distributed machines.
"Vendors should make sure they are testing patches extensively before pushing them out the door," he says. "This is much easier for open source technologies than it is for closed source companies like Microsoft, which tends to have longer lead times on patches and fixes."
Yet industry watchers say the problems with patching today and going forward will have more to do with user environments than with vendor updates. They warn that while patch management technology can be considered mature, as environments evolve to include more virtualization and complex application infrastructures, patching will need to grow up. And in turn, vendors such as Altiris (now part of Symantec), BigFix, CA, St. Bernard, PatchLink and Shavlik will need to bring support for virtualization and other technologies into their tools to adequately patch customer environments.
"It is a relatively mature technology, but that doesn't mean it's under control. Patching in areas such as virtualization remains considerably immature at this point; server and desktop virtualization are throwing the old rules for patching out the window," says Andi Mann, research director at EMA.
According to Mann, virtualization introduces complexity as well as exponentially more machines to be patched in the same amount of time. This could cause IT managers to hasten the patch testing process, which would ultimately cause configuration conflicts on production machines. "Testing is a critical element of patching, and with immediate threats like zero-day attacks and the proliferation of virtual machines being certain all the updates will work together and protect the environment is more difficult," Mann says.
Add to that dependency patching, says Jasmine Noel, principal analyst at Ptak, Noel & Associates, and patching could become a new challenge for IT managers. As environments get more complex and the number of vendors distributing patches grows, she says, the layers of testing and distribution of updates will break the old approach to patching for many IT managers.
"What still needs work is sequencing patches from multiple vendors because of all the infrastructure layers -- hardware, operating system on the physical box, virtualization software and the virtual machines (operating system and application stack) -- so what is the right sequence to install your HP, Microsoft and Oracle patches that all came out last Tuesday?" she asks.