One of the problems I hear about all the time, from analysts and vendors alike, is that of rogue mobile clients. Youve got perimeter defenses on your corporate network worthy of the Maginot Line, but mobile users can still travel, connect to the Internet on the road, pick up a germ on their notebook and release it when they log back in at the office.
There are many products and good practices to prevent this, but one of the trends I see developing is that of qualifying clients to connect to the network. When attempting to connect, the client is first queried for whether it meets any of a number of security parameters: Is it running a recent version of antivirus software? Are the definitions up to date? A firewall? Are critical Windows patches applied?
Ive seen this approach taken by a number of vendors, usually with a proprietary approach. The Symantec Series 300 small business security appliances can be set to enforce use of most Symantec antivirus clients. Another appliance vendor Ive been talking with has similar support for McAfees clients in beta.
But the most interesting and comprehensive approach to this is the Cisco Network Admission Control program (NAC). Its still not exactly clear what NAC will turn out to be, but at a minimum it will be a powerful tool for Cisco customers. Taken all the way, it could turn into a general framework for client security policy enforcement. It all depends on how much Cisco decides to open up.
I talked to David King, a director in Ciscos VPN and security business, about NAC. According to King, Cisco observed about a year and a half ago that conventional network defenses, such as firewalls and antivirus gateways, were not successful enough. (Gee, you must need an expensive consultant to tell you that; never would have occurred to me.)
Networks are porous for lots of reasons, not least of which is the traveling employee mentioned above. NAC helps to create what Cisco calls the “self-defending network.”
Next page: How Ciscos NAC program works.
How NAC Works
There are two main parts to NAC, the first being a piece of software that runs on the client (or, more generally, “endpoint,” because it could be a server, too, or something unconventional) called the Cisco Trust Agent or CTA. At the back end is a policy management system called the Cisco Access Control Server (ACS). Cisco gives away the CTA, and the ACS is already widely used for various 802.1x-based authentication functions.
Cisco approached the big client security companies (Symantec Corp., Trend Micro Inc., Network Associates Inc.s McAfee Security) and got their cooperation. These three probably constitute more than 90 percent of the corporate antivirus market.
The protocol between the CTA and client software allows it to check the status not only of the antivirus software and definitions, but also firewalls and the operating system version and patches. Depending on what is found, the administrator can set policies to lock out the user or place them in a restrictive environment.
King says Cisco could open up the APIs for communication between the client software and the CTA, as well as the APIs between the ACS and other vendors policy managers, and that would be great.
If they opened up the protocols between the CTA and the ACS, that would be even better, although I can see why Cisco would hesitate to do it, since it would open up the possibility of third-party implementations of NAC.
Consider that if NAC caught on, there would be a lot of pressure for a built-in Linux client; lots of people would want this to be a source code version. Cisco is still deciding on how to proceed; I think it would do well to be aggressive in opening it up. It would accelerate general acceptance of the system, and that would only be good for Cisco, which plans to put NAC in everything from routers and switches to wireless access points.
Eventually Id even like to see ISPs implement something like this. Id pay extra for an ISP that did have such rules, but I wouldnt put any money down that they will. ISPs seem loath to lose even their infected customers.
Its true that new attacks come along, and NAC wouldnt be able to stop most of them, but the vast majority of the problem out there comes from old attacks, or at least from those detected by the current definitions of any major AV product. With NAC or something like it in place, networks will be able to defend themselves against users who dont know enough to protect themselves.
Security Center Editor Larry Seltzer has worked in and written about the computer industry since 1983.
More from Larry Seltzer