Networking's greatest debates in Security
- 29 October, 2007 08:05
A look at the all time greatest controversies in the history of the networking industry.
Immediate flaw alerts vs. Disclosing with patches
What's safer, knowing there's a gaping hole that can be exploited in a software product even when there is no patch for it, or being told about the gaping hole once there is a patch?
That debate, heard since the dawn of software, pits the tell-all crowd arguing for "full disclosure" against those who argue for "responsible disclosure," a philosophy favoring greater discretion about software vulnerabilities in the hope that malicious hackers won't benefit from too much information.
But that assumes they don't already know anyway. And if the hackers know, then is it just the good folks who are in the dark? Such have been the powerful arguments on both sides, which grew louder in the 1990s as Microsoft Windows settled in for a long stay on the desktop and server, giving "script kiddies" armed with automated attack tools the ability to hit a lot with little effort over the Internet. It didn't help that Microsoft in the early days was in a blissful state of near-complete denial about software holes.
At the same time, security research was accelerating, with brash young firms like eEye Digital Security (founded in 1998) discovering vulnerability after vulnerability in Windows, and at the time, arguing for full discovery. Then the real impact of software vulnerability hit home for the entire world when the crippling computer worm named Code Red ripped across the Internet in 2001, exploiting a vulnerability in unpatched Microsoft ISS Web servers.
Although a server patch had been available for a month that could have stopped Code Red if applied to servers, the topic of disclosure grew ever more shrill as some accused eEye of revealing too much about Windows flaws.
In an attempt to find balance in the debate, a group calling itself the Organization for Internet Safety was founded in 2002 by Microsoft and others in the industry to come up with guidelines for responsible disclosure of software flaws. Last updated in 2004, the OIS guidelines say someone discovering a software flaw should discretely share that information only with the software vendor involved, allowing a minimum of 30 days to correct the problem.
But since then, the argument has only gotten more muddied as a thriving industry in the last few years has sprung up for selling information about vulnerabilities directly to security firms, which then market the vulnerability data to subscribers.
Some individuals who once backed the OIS guidelines say they're antiquated and only useful for protecting software vendors. "The OIS standards were a valiant effort, but in the end the OIS was designed to help vendors manage things on their end," says Terri Forslof, who helped craft the OIS guidelines when working in Microsoft's security-response center but joined a security firm re-selling vulnerability research.
Still, others vehemently disagree, saying responsible disclosure in which vulnerability research is shared first privately with the software vendor is ethical, while selling it to subscribers is not. "They're brokering information that makes the world less safe," says Kris Lamb, director of the X-Force research development at IBM's Internet Security Systems division. -Ellen Messmer
IDS vs. IPS
A firestorm of controversy exploded four years ago when consulting firm Gartner declared that intrusion-detection systems that passively monitor for malicious traffic would be "dead" by 2005, a dinosaur wiped out by intrusion-prevention systems that proactively block bad traffic.
Buying an IDS to monitor unwanted traffic is a waste of time and money, Gartner stated, urging enterprise managers to start buying in-line IPS products and step up to the plate and block the attack traffic comin' at 'em, primarily from the Internet.
Blocking the bad traffic with an in-line IPS opened the possibility of mistakenly blocking good traffic, too, yelped IDS proponents.
IPS products in 2003 were mainly in their infancy and their accuracy deeply suspect. IDS - the most well-known and popular being open source Snort created by Martin Roesch in 1998 - was a known quantity. Sure, IDS had its drawbacks, sometimes generated false positive and negatives, and most people didn't really know what to do with the massive amount of information netted in the monitoring process.
But Gartner saying IDS is dead?
"I find the logic behind their conclusions significantly flawed and their recommendations incomprehensible," was the response at the time from Roesch, CTO at Sourcefire, founded in 2001 to commercialize Snort. "To be fair, Gartner's concerns have some basis in fact," he conceded, adding, "Undoubtedly, IDS must continue to evolve in order to fully realize its potential."
Today, the issue is largely a moot point as IPS products on the market - which typically rely on IDS detection techniques to flag a problem - tend to operate in a mixed mode, allowing managers to boldly block malicious traffic or passively monitor, or both, depending on the configuration. Security vendors are often coy about breaking out figures on IDS and IPS, but IDC believes IPS began overtaking IDS in 2005. Continuous testing by independent sources helps with determining strengths and weaknesses in IPS. -Ellen Messmer
IPSec vs. SSL VPNs
When IP VPNs came on the scene in the late 1990s IPSec quickly established itself as the standard to provide secure network-layer connectivity over insecure IP networks, typically the Internet.
The appeal was obvious: it is less expensive to buy Internet access and make WAN connections over it than to buy dedicated circuits or a frame relay or MPLS service.
But IPSec is complex. The more sites that connect to each other, the more secure links or tunnels need to be defined and maintained. If IPSec is used for remote access, it requires software on every remote machine that must be installed and maintained.
Then SSL VPNs entered the scene offering application-layer secure access over the Internet using capabilities common to most browsers. The implication was that businesses interested in remote-access VPNs no longer needed to distribute and maintain client software on the remote machines.
The limitation of SSL was that the browsers could access only Web-based applications, but this challenge was met by Webifying non-Web applications or pushing Java or Active X SSL VPN agents to the remote machines on the fly. These plug-ins gave the remote computers the ability to create network layer connections comparable to IPSec, but without having to distribute dedicated VPN client software.
As a result, SSL VPNs are making great headway against IPSec VPNs for remote access and seem likely to win out in the end.
IPSec is still the preferred method of site-to-site VPNs because either technology requires a gateway anyway, IPSec is better established in this arena and many SSL vendors don't even offer site-to-site connections. For site-to-site, IPSec carries the day.
Perimeter security vs. inside security
When businesses began hooking up to the Internet in earnest in the late 1980s, it was with a sense of trepidation and awe, knowing an unprecedented public interaction was commencing. In the hope of holding dangers at bay, the bastion firewall emerged as the fortress guard, thanks to technology innovators such as Marcus Ranum and Bill Cheswick. Early commercial firewalls, including Digital Equipment Corp.'s SEAL, meant enterprises would no longer have to roll their own.
The perimeter firewall has become a fixture, the point of demarcation where specialists lavish attention on complex security rules to define permitted inbound and outbound traffic. But 20 years later, the role of the Internet firewall and similar perimeter defense has come under sharp question by a growing number of security managers who base their arguments on one simple point: the perimeter has disappeared.
The demands of e-commerce to access internal systems, collaboration with outsourcing partners, the mobile laptops and computer-based handhelds carried by business people to the ends of the earth - these all contribute to the "disruptive change," argues Paul Simmonds, chief information security officer at U.K.-based chemicals and paint manufacturer ICI.
"Your security perimeter is disappearing," notes Simmonds, energetic supporter of the Jericho Forum, the group founded by corporate information security managers in early 2004 to encourage the development of more innovative data-centric approaches to enterprise security that reflect today's malleable business situation. "What we're architecting at the Jericho Forum is not an individual solution, a single fix. We call it a collaboration-oriented architecture."
Jericho Forum now has about 45 members, mostly large European firms but with more U.S.-based ones joining these days.
One of Jericho Forum's favored terms is "de-perimeterization" (the British spell it with an 's' not a 'z') and while the group doesn't specifically advocate doing away with perimeter firewalls, its critique of them as a barrier to e-commerce has at times elicited strong opposing opinions that the group's views are wrong-headed, misguided or naive.
"At best, Jericho will help raise awareness of the usefulness of a defense-in-depth network security strategy," stated Joel Snyder, senior partner at Opus One and a member of Network World's Lab Alliance, writing about the group two years ago. "More likely the Forum will end up on the scrap heap of unrealized ideas and wasted effort." Snyder says his opinion that some of the Forum's thinking is "moronic" is no different today.
Firewall innovator of legend, Cheswick, lead member of technical staff at AT&T Research, acknowledges it's appealing to consider a world where corporate security doesn't rely on perimeter defense. But in his keynote presentation at the Jericho Forum meeting in New York last September, Cheswick said the limitation in foregoing perimeter defense is that "you won't stop a [distributed denial-of-service] attack, so we may still need a walled garden."
Nonetheless, the Jericho Forum soldiers on with its work to convince enterprises and vendors alike to think outside the perimeter box. "De-perimeterization for most corporations is a fact of life," Simmonds points out. "For most corporations, it's happening whether you like it or not." -Ellen Messmer
Read Networking's greatest debates in Software
Read Newtorking's greatest debates in the Data Center
Read Networking's greatest debates in Management
Read Networking's greatest debates in LANs and WANs