Feature-Rich Firewalls Needed
The Jericho Forum predicts a day when it will be unnecessary to build network perimeter defenses with firewalls, but in the meantime corporate network-security experts need to find alternatives to address the shortcomings that are pushing firewalls toward extinction.
They need to look for equipment that filters at the application layer and supplemental products that proxy encrypted traffic so it can be inspected, experts say.
"A next-generation firewall needs to look within traffic streams and determine whether this is the traffic I expected," says Rob Whiteley, an analyst with Forrester Research. The key to protection is peering deep into packets to decide what poses a threat and what doesn't, not merely on what ports it uses, he says.
Firewall vendors already recognize this and have incorporated deep inspection of packets that probe to the application layer to determine the nature of traffic and look for anomalies that can signal malicious behavior, he says.
Most businesses leave open port 80 for HTTP and port 443 for SSL, and these protocols are used to transport a wide variety of traffic and applications, legitimate and otherwise, says Greg Young, an analyst with Gartner. With these ports unguarded, traditional firewalls allow in more potentially damaging traffic. Implementing Web-based networking accelerates the problem.
SSL traffic, which is growing rapidly as a percentage of corporate traffic, poses a particular problem because there is no way for a firewall to decrypt it to find out what it contains, Young says. "The fact is this is a real blind spot. It's a dirty little secret that a large percentage of traffic is not being inspected," he says.
The only way around the problem is to terminate the SSL, decrypt it and inspect it, which effectively creates a man-in-the-middle attack, albeit an authorized one, he says. Nevertheless, it still breaks the security model that encryption secures traffic end-to-end as a way to thwart man-in-the-middle attacks, says Young.
Businesses have to weigh the chances of damage from traffic that comes in encrypted vs. damage that could come about by breaking terminating the SSL session before its destination. "The odds of bad things coming through the SSL sessions are higher," he says.
Proxying SSL is an approach being taken by Blue Coat and Palo Alto Networks, among others. Palo Alto says its gear can be configured to permit some SSL traffic without proxying, such as personal online banking as determined by packet headers.
The gear also distinguishes among different flavors of similar traffic such as e-mail, and can impose fine-grained rules. For instance it could allow both Yahoo Mail and Outlook mail, but blocking attachments from Yahoo Mail.
SanDisk is trialing Palo Alto gear in monitoring mode, and finds the visibility it gives in to application use on the network to be valuable, says Justin Smith, the company's network architect. When the product is installed inline in a few weeks, he says it will also enforce rules based on users, not just machines. So if a user logs in from a different switch port or VLAN, the firewall will enforce the same user-based rules, Smith says.
He says the device recognizes applications themselves, not just what port they use so if an unauthorized application switches ports, the gear can still block it.
Another alternative is to consider moving the location of deep-inspection firewall protection, Whitelely says. Personal firewalls can be installed on desktops and laptops, and application layer gateways can filter traffic as it approaches critical network assets.
"The next-generation firewall could be the last piece of network infrastructure before servers," Whitely says. This could be part of the function of load-balancing, application-accelerating gear such as that made by Citrix, F5, Imperva and others.
One user of a firewall with application-aware intrusion prevention capabilities says the transition from traditional firewalls should be made slowly. Bimba Manufacturing, a pneumatic equipment maker in Monee, Ill., has installed a Secure Computing Sidewinder firewall to help protect its network as more Web applications roll out, says Matt Nantais, system analyst for the firm.
It is still using IPS features of its Secure Computing Sidewinder firewall in monitoring mode until the company can determine the extent of false positives and how to minimize them. The company hopes that the IPS when turned on will drop attack traffic it identifies as potentially harmful to servers or the network.
As businesses consider moving to next-generation firewalls, they should weigh certain key factors, Young says.
These include, aligning their replacement schedule with the replacement schedules for other gear that now comes as part of some firewalls in bundles called unified threat management (UTM) devices. The devices may incorporate antivirus, antispam and content filtering.
If they align their firewall, IPS, URL filtering and Web antivirus refresh times they may gain option of merging them into fewer devices. So customers should check out what their current vendors offer. "You might not have to deploy a new appliance," he says.
Customers should consider their current firewall vendor for supplemental features, Youg says. Their current vendors will integrate these added protections with the firewall, eliminating the work of making sure the firewall is compatible. "You certainly don't want to break the firewall when you implement IPS," Young says.
He also recommends weighing the cost of retraining staff to administer a new firewall. It may be worthwhile sticking with the current vendor, he says.