Crossing VoIP's security minefield
VoIP creates a new set of challenges.
By Jeff Prince, Network World | Network World US | Published: 14:00, 04 December 2007
I'm hearing more about new kinds of attacks on the LAN, such as voice over IP (VoIP) attacks or using printers as a source of attack. How can LAN security help me prevent such events?
It's absolutely true that these kinds of attacks are on the rise. In fact, the SANS Institute recently listed client-side attacks as one of today's most critical vulnerabilities. While it may be foolhardy for any of us to think we can fully prevent such attacks, you certainly have some strong options for mitigating against them in your enterprise.
One of the first steps to take is to implement an authentication scheme in your LAN that encompasses devices as well as users. If you're pursuing something like 802.1X, it won't be sufficient because phones, printers, medical devices, robotics, and other devices for the most part will not be able to support the required 802.1X supplicant.
You need a way to ensure that you're aware of every non-user device plugging into the network and that you will know what kind of device it is. Look for an authentication approach that lets you whitelist your specific known devices or, even better, helps you identify those devices automatically by using reverse DNS to associate a device name and type with the device.
Next, you'll need a means for placing these non-user devices into a role and assigning access rights to that role. For example, you could define a printer role that would apply to all the printers and print servers in your environment. As for access rights, you could specify that printers could communicate only with the print server and that all user devices are able to communicate only with the print server. With this kind of policy, you'd prevent direct communications between a user device and a printer.
Similarly on the VoIP side, you could assign VoIP phones to a VoIP role and define that those VoIP phones can communicate only with the call manager. You can even go beyond this kind of zone-based protection with application-based policies. For instance, you could say that devices in the VoIP role should be communicating only in SIP, H.323, or SKINNY, for example, to further protect against a data-based attack.
This kind of zoning and partitioning is very helpful in preventing phones, printers, or other devices from being used a launching point for an attack. A printer that had been compromised and had vulnerability scanning software loaded onto it, for example, now would not be able to reach out to all your network devices looking for open ports. And a VoIP phone could not be used to launch an attack on other servers or end user machines, and with application protection, it couldn't even attack the call manager using a data protocol.
In what form can you get such LAN security protection? You have a couple options. Next-generation LAN switches, with authentication beyond 802.1X and the ability to apply policy-based access control to users and devices, are a great way to get this capability built directly into your LAN. If you're not looking at a switch upgrade yet, then consider security appliances that have the ability to authenticate users as well as devices, assign devices to roles automatically, and apply policy-based controls by zone and application.
Whether you choose an access switch or an appliance, it's critical that the protection is applied right at the user edge of the LAN. This location is essential for mitigating against these kinds of client-based attacks. Otherwise, you won't have the tools to stop the traffic flows right where they start.
Jeff Prince is CTO, ConSentry Networks.