If you think all IDS/IPS signatures are created equal, think again.

The quality of an IDS or IPS signature may be what stands between you and a targeted attack, says researcher Simple Nomad, a.k.a. Mark Loveless, who will demonstrate gaping holes he's found in IDS and IPS products next week at IT Security World in San Francisco.

"A lot of people say the whole model of IDS/IPS is broken," says Loveless, who will present his latest research on bypassing IDS and IPS systems. "But the problem here boils down to poor signature-writing in some instances... Good signature-writing is effective."

One big problem is some signatures for filtering exploits are written to the publicly disclosed exploit, rather than the underlying vulnerability, he says. So if a known exploit's payload code is 4,096 bytes, for instance, the IDS or IPS signature would "look" for that characteristic to filter out the exploit. But a clever attacker could merely alter the size of the exploit's payload to, say, 5,000 bytes to avoid detection by the IDS or IPS.

"That happens a lot -- signature-writers write against the exploit as opposed to the vulnerability," Loveless says, and since most vendors outsource at least some, if not all, of their IDS/IPS signatures, this can provide attackers an easy way in.

Loveless, senior security architect for Vernier Networks, which makes a network access control (NAC) product that does intrusion prevention with a combination of signatures and anomaly detection, says part of his job is to check the quality of signatures that Vernier outsources. "Most of the time they come in fine, but sometimes a vendor will send five signatures and I may have to write an additional three signatures to provide the proper coverage, as well as edit their five signatures."

IDS/IPS evasion research isn't new -- Metasploit creator HD Moore demonstrated at Black Hat USA 2006 some application-level exploits that zip right past an IPS. (See IDS/IPS: Too Many Holes?)

Loveless' research, meanwhile, focuses more on how attackers can use weaknesses in signatures as well as how they can case, or scan, the perimeter to detect which brand of IDS or IPS is sitting on the victim's network, and then use that intelligence to wage the ultimate attack.

In the first wave -- reconnaissance -- the attacker can determine whether the device is an IPS versus an IDS if he can access "XYZ" before a first-stage attack, but not afterward, Loveless says. Delays in blocking the attack may be a clue that an admin is reading his IDS logs -- thus it's an IDS.

Loveless says he will demonstrate how an attacker can then manually "fingerprint" the brand of IDS or IPS, and then adjust the attack to take advantage of that product's weaknesses. "If you have some idea of what brand of IPS covers this vulnerability or that one, you can fingerprint it and say this looks like Brand X or Brand Y."

Fingerprinting specific products is more difficult for an attacker to do than probing the perimeter to determine if it's an IDS or IPS, however, he says.

But even the most airtight of IDSes or IPSes won't stop a determined attacker. "Ideally, every probe that came in, [the IDS or IPS] would block," Loveless says. "But then that tells the attacker he shouldn't go through the network... Instead he should try the WAP [wireless access point]."

Have a comment on this story? Please click "Discuss" below. If you'd like to contact Dark Reading's editors directly, send us a message.