Despite spending millions of dollars on technology and countless hours on defense, large enterprises continue to be hacked and infected at an unprecedented pace. Clearly, conventional approaches to security aren't getting the job done.

With this harsh reality in mind, many security professionals -- and the vendors that serve them -- have begun some slow, but fundamental, changes in the way they view their approaches to the IT security problem. And as a result of these changes, tomorrow's security organizations in large enterprises -- particularly the security operations center (SOC) -- could work very differently than they do today, experts say.

"People in the SOC need ways to react faster and better -- they need ways to improve the efficiency of what they do," says Rich Mogull, founder of Securosis, a security consulting firm. "They need ways to reduce the amount of time between the onset of an attack and the time it's stopped or remediated."

To achieve this efficiency, enterprises may need to undertake some tectonic shifts in the way they think about security and the way they spend their days. Let's look at some of those changes in philosophy and how they might affect the future activities of tomorrow's SOC.

Tomorrow's SOC will spend more time on security analytics and less time on perimeter defense.
Traditional notions of "defending the walls" of the enterprise are crumbling, experts say. As users become more mobile and businesses become more interdependent, the "security perimeter" of a given organization is becoming increasingly harder to define -- and nearly impossible to defend.

"In 2007, it's estimated that there were about 500 million network-connected devices on the planet. By the end of 2010, that number will grow to 35 billion -- that's five devices for every person on Earth," says Don Proctor, senior vice president in the office of the chairman and CEO at Cisco Systems. "We can't secure them by patching them all at the endpoint. We have to say goodbye to the perimeter."

In fact, some security professionals -- even some vendors -- are rejecting the fundamental notion that the perimeter can be completely defended. The new assumption: The enterprise will be compromised, and probably already has been.

"At this point, the security team can have no confidence that a given host has not been compromised -- the bad guys are already inside your environment," says Amit Yoran, CEO of security vendor NetWitness and a former White House cybersecurity adviser. "All of the threats that really matter are already inside the network."

While not all security experts agree on this philosophy, most agree that tomorrow's security teams will have to spend at least as much time analyzing logs, events, and incidents as they currently do on building perimeter defenses. That means more focus on security analytics, forensics, and incident response.

"Over time, the people in the SOC will find that they're spending more time as data analysts, rather than security analysts," says Joe Gottlieb, CEO of SenSage, a maker of security information and event management (SIEM) tools. "They'll be doing a lot more data mining to find the source of a problem. They'll be more focused on, 'You've been hacked. Now what?'"

Tomorrow's SOC will spend more time identifying new, unknown threats and less time blacklisting known threats.
Even the antivirus vendors that invented the concept now agree: The notion of building security defenses around the "signatures" of known attacks is not an effective long-term solution.

"Ten years ago, we were identifying five or 10 new viruses each week that needed to be blacklisted," said Gerry Egan, director of Symantec's Technology and Response unit, at the recent unveiling of Ubiquity, the company's new reputation-based security tool. "Now we're identifying 10,000 to 15,000 new signatures every day. The old, signature-based model has become a bit long in the tooth."

While signature-based technology will continue to be part of enterprise security strategies, the analysts in tomorrow's SOC will spend more time seeking out changes in network and systems behavior that might indicate new attacks, experts say. Emerging technologies, such as Symantec's Ubiquity, Dasient's Web Anti-Malware, and FireEye's Malware Protection System, could improve detection of zero-day threats because they seek to identify changes in behavior and reputation, rather than focusing on known threats.

"With the growth and evolution of malware today, if you attempt to defend against attacks solely via signatures, you are doomed to fail," says Neil Daswani, co-founder and CTO of Dasient. "We have to shift our thinking from focusing on what the code looks like to what the code does."

This means the staff of the next-generation SOC will likely spend more of its time in malware analysis -- even malware research -- than ever before, experts say. Mogull suggests tomorrow's SOC will need to develop its own combination of correlated activity sets and signature feeds, essentially creating a threat analysis environment that is tailored to the specific threats, risks, and sensitivities of the business.

The next-generation SOC will also need a better process for quickly analyzing behavioral data that might indicate new threats and escalating it to the top of the security team's priority list, Mogull says. Many SOCs will develop customized scripts and user interfaces that help automate the escalation process and speed the analysis and resolution of potential security problems, he says.

Tomorrow's SOC will spend less time aggregating events and more time doing proactive monitoring and intelligent correlation of security data.
For years, the SOC has been defined by SIEM tools that collect information about security-related "events" on the network and consolidate that data onto a single monitoring screen. Experts agree these tools aren't going away, but mostm-- even the makers of SIEM systems themselves -- also agree that security monitoring in the next-generation SOC must become smarter than it is today.

"The current security monitoring environment gives you a useful approximation of what's going on," says Gottlieb, of SIEM vendor SenSage. "Most of the [monitoring] technology can't accept data from any source. And even so, there's a huge amount of data in logs and SIEM systems to go through. It's hard to isolate the signal from the noise."

And the data analysis problem isn't getting any easier, Mogull observes. "There's a ton of data in SIEM, but in the end it's only log-level data," he says. "What will be needed down the road is the ability to do network-level analysis, even packet-level analysis, and that's not all going to happen in one single system."

The next-generation SOC will require new technologies that can correlate data from a wide variety of security systems and help present it in a way that enables analysts to more quickly dig through the haystacks of security information and find the needles of data that could indicate a breach. Realistically, the SOC will get smarter not by consolidating on a few security systems and applications, but by improving the ability of analysts to pull relevant data from many different systems and correlate it to find the source of a threat.

"What's really needed is the kind of event correlation that allows you to baseline your network so that you can understand what 'normal' is," says Cisco's Proctor. "If a laptop that previously never communicated outside the building suddenly starts sending data to Brazil, you want to be able to understand when it started and what data is involved." Eventually, that correlation might extend even to physical systems, so that the SOC can recognize patterns of usage on secure doors and surveillance cameras as well, he says.

In addition, the next-generation SOC staff of the future will do more proactive monitoring and less reactive monitoring, according to some experts. While SIEM and event correlation tools may help identify the source of a breach more quickly, they still don't stop the attack from happening, they note.

"I think a vision that says we're already hacked is defeatist -- and a little bit bizarre," says Steve Dauber, vice president of marketing at RedSeal Systems, which makes tools designed to test vulnerabilities in security policy and measure the security posture of an organization. "If you look at the data from the most recent Verizon Breach Investigation Report, you'll see that most breaches occur because companies fail to take some fairly simple steps to secure their systems. What they need are more ways to find those problems proactively, before the breach occurs, rather than analyzing them post-hack."

Avishai Wool, co-founder and CTO of security configuration management tool vendor AlgoSec, agrees. "I don't agree that the perimeter is dead," he says. "There are many ways to more effectively prevent attacks, but we need a lot more automation and a lot better tools. Mortals cannot configure [virtual private networks], which is what a lot of enterprises are trying to do today."

Both RedSeal and AlgoSec are offering tools that enable enterprises to more effectively evaluate the configuration of firewalls and other security systems to help find vulnerabilities and test them against corporate security policies. This proactive analysis and testing, along with traditional vulnerability scanning, is designed to help companies find the holes in their security systems before the bad guys can exploit them.

Mogull suggests that some existing tools, such as data leak prevention (DLP) technology, could also help in identifying potential leaks and preventing sensitive data from leaving the corporate domain. "DLP gets criticism from security people sometimes because there are ways around it," he says. "But the fact is that DLP can be a big help in egress filtering and identifying advanced persistent threats. There are a lot of bad actors involved [in breaches], and not all of them are smart."

Tomorrow's SOC will spend more time working with security service providers and less time in "do it yourself" activities.
By almost all measures, the volume and complexity of security threats is growing at a breakneck pace. In figures releasef today, Dasient is reporting that there are more than 1.2 million malware-infected websites on the Internet, more than twice as many as there were a year ago. Statistics released last week by CompTIA indicate that nearly two-thirds of enterprises have experienced at least one breach in the past 12 months.

Yet the human and budget resources devoted to IT security are barely increasing at all. In a study released in June, Gartner reported that security spending decreased from 6 percent of the overall IT budget in 2009 to 5 percent this year. Clearly, tomorrow's SOC cannot hope to keep up with the growing threat environment solely by adding more in-house staff and technology.

For this reason, many enterprises are looking toward security services to help them handle some elements of their defense, experts say. While software-as-a-service offerings from major security vendors, such as RSA, Cisco and Symantec, continue to increase in popularity, smaller startups, such as Immunet, FireEye, and Invincea, are building their businesses around services, rather than software.

"Companies are seeing that deploying client-side technology and constantly patching doesn't always work," says Daswani of Dasient, another service-based provider. "I'm not convinced that all of the right endpoint security tools are even available yet. We're seeing some real efficiencies in looking at these problems from the server side."

Research from IDC indicates that the security services market will grow from $32.3 billion worldwide this year to $44.1 billion in 2010. While many of these services will be used by small and midsize enterprises, SOCs in large enterprises will also be looking to leverage the data that is collected by security vendors on malware and other threats, experts say.

But that doesn't mean the expertise of the SOC staff will become less important, experts say. In fact, most experts agree the next-generation security analyst will have to be smarter than ever -- not only about current threats, but about their potential impact on his or her specific business environment.

"The industry's hope to staff the SOC with cheap labor is over," SenSage's Gottlieb quipped. "Security is not going to get simpler. The security staff of the future is going to need expertise not only about the domain they're defending, but also contextual expertise to determine what combinations of events might present a threat. On top of that, they're going to need analytical expertise so that they can determine the source of the threat -- and how to stop it."

Have a comment on this story? Please click "Discuss" below. If you'd like to contact Dark Reading's editors directly, send us a message.