Menu
Extend perimeter defence to application layer

Extend perimeter defence to application layer

I am the biggest defence-in-depth supporter there is, but deep defences don't obviate the need to protect the perimeter. This year, we are going to see a lot of action from security vendors trying to provide better and stronger defences at the border from the growing plagues of viruses, spyware, malware and phishing attacks.

The firewall gang has jumped on the idea of stronger perimeter defences with their own fairly meaningless buzzword, unified threat management (UTM). The core is that firewall protection can go beyond allowing or blocking particular services. Instead, firewalls can scan traffic for viruses, handle content filtering, do basic intrusion prevention and watch for signs of spyware.

Every firewall vendor has a different idea of what goes in UTM, but there is a common thread: application-layer inspection. If you move up the communications stack from the most basic features of a firewall, you will see that application-layer inspection is the last word in perimeter protection. The concept is to not merely look for signatures or protocol anomalies but actually understand what is happening at the application layer and use that knowledge as part of a protection policy.

Firewalls aren't the only place where perimeter defences are being bulked up. Intrusion-prevention systems, typically placed at the perimeter, often reach up into the application layer. Web proxy servers also will see a lot of action in 2006, especially following Blue Coat Systems' success. Expect to see a big push from firewall vendors for their UTM features and from other security vendors for application-aware perimeter defences. When evaluating these high-pressure marketing pitches, keep three guidelines in mind:

  • UTM is a buzzword. You can't compare UTM products side by side, because no two do the same thing. Even when the labels look the same, you'll find differences. Gateway antivirus is a common feature of perimeter devices, but products vary wildly in the protocols they examine and how deeply they look into each one. For example, you may find that an antivirus product scans HTTP traffic on Port 80 but not on Port 8080, or that it cannot find viruses that enter via Webmail services - such as Yahoo, MSN and Hotmail - because of the way the browser views messages.
  • Performance is critical. It is very difficult to test performance reliably and fairly as you move up the network stack to the application layer. Security vendors may not even know exactly how fast their products can work. In any case, if you are pushing UTM features into an existing firewall, make sure you have a lot of headroom - at least 90 per cent of CPU free - for the load before even thinking of using existing hardware. You may want to replace that old firewall anyway. Unlike a good cheese, firewalls get softer as they get older.
  • Subscriptions are expensive. We are moving from a capital-cost model in security to a subscription model. Make sure you consider support and software costs, which can easily exceed capital costs even in the first year, when adding application-layer security.

Snyder, a Network World Test Alliance partner, is a senior partner at Opus One in Arizona. He can be reached at Joel.Snyder@opus1.com.


Follow Us

Join the newsletter!

Error: Please check your email address.
Show Comments