Orignially posted on http://www.nta-monitor.com
…10,000 username guesses in 2 minutes 30 seconds…
During the course of regular testing, NTA Monitor have discovered two serious flaws in Checkpoint Firewall-1, giving rise to both username guessing and sniffing issues.
Firstly, affected versions permit remote users to determine if a Firewall username is valid without having to know the associated password, enabling hackers to guess valid usernames using a dictionary attack. In tests of the flaw conducted by NTA Monitor, it took 2 minutes 30 seconds to check 10,000 usernames at a rate of 67 guesses per second using only 10% of a 2Mbps leased line. The guessing rate is mostly limited to by the Firewall CPU rather than by the Internet link speed. In effect, this means that companies using a hi-spec firewall server increase the speed at which an attacker can guess passwords.
In addition, VPN usernames are passed in the clear without encryption, allowing anyone who is able to sniff network traffic between VPN clients and the Firewall to observe usernames in transit. The flaws exploit the Internet Key Exchange (IKE) encryption scheme and affect all Checkpoint Firewall-1 systems of 4.0 or above.
This leaves the back door to the enterprise wide open to hackers. The biggest problem is that it is not necessary to send a password to obtain a reply from the Firewall. Given that both users and system administrators often chose weak passwords, it is likely that any attacker will be able to guess at least one password and thus gain access to the VPN - and from there most configurations easily allow full access to the company's resources.
The correct approach would be to wait until both username and password are supplied, and if either is incorrect, send a generic error message. We were surprised to find this flaw when this is standard security practice in many other authentication mechanisms, including Unix logon.