VoIP hacker says provider networks are easy pickings
- 19 August, 2007 22:00
A combination of simple dictionary and brute-force attacks in combination with Google-hacking enabled a criminal pair to break into VoIP-provider networks and steal US$1 million (NZ$1.3 million) worth of voice minutes, says one of the duo who has pleaded guilty to his crimes.
Had his victims observed security basics, most of the attacks would have been unsuccessful, says Robert Moore, the 23-year-old hacker, who has been sentenced to two years in federal prison and fined US$150,000.
In the interest of corporate telecom executives that want to lock down their VoIP networks, Moore reveals his methods in a podcast interview with Telecom Junkies at thevoicereport.com.
Moore says he wrote generic software to run brute-force attacks against Cisco XM routers and Quintum Tenor voice gateways to gain access to them so he could route calls through them. These devices were located in business networks, and calls were routed through them to mask that they came from gear owned by the mastermind behind Moore’s activity, Edwin Pena.
Pena was arrested last year along with Moore, but after posting bail fled the country and has not been caught.
Moore also conducted brute-force attacks against service provider networks in order to discover valid prefixes to let calls into their networks.
He designed software to generate 400 prefixes per second against the carrier gear, scanning all the combinations between 000 and 999 randomly to throw off intrusion-detection systems (IDS) that might pick up a sequential attack, Moore says. The attacks were made against VoIP gateways using the H.323 signalling protocol, but not those using SIP, he says.
The pair also scanned known corporate IP addresses for machines that might be vulnerable to their attacks, Moore says. Pena purchased a 2GB database of corporate IP addresses and their subnet ranges for $800, he says.
“The way we got into them is that most of the telecom administrators were using the most basic password — Cisco, Cisco or admin, and admin. They weren’t hardening their boxes at all,” Moore says.
Pena and Moore found many devices on the internet with exposed SNMP ports that allowed probing for private information. “There were various object identifiers in the management database that would allow you to see critical information on a Cisco [router], like maybe [the] gateway where it’s routing to so we would know where to choose our target,” he says.
The object identifiers also helped them identify exactly what make and model machine they had found, and they used that information to research vulnerabilities those machines are known to have so they could exploit them, he says.
Moore also wrote search strings that he fed into Google seeking exposed web interfaces on devices, and that proved fruitful as well.
“It was really easy actually to launch these things from Google to find these peoples’ switches,” Moore says.
Dan York, a director of the VoIP Security Alliance who participated in the Moore interview, says standard best practices would thwart Moore’s and Pena’s tactics. “The attacks were relatively simple attacks that could have been prevented by IT Security 101,” he says.
For instance, brute-force prefix searches could be headed off by limiting the number of log-on attempts and timing out attempts after a set period. Closing up unused ports of internet-exposed devices would reduce the chances of being probed, he says.
York also suggests businesses use attack tools published by VoIPSA to test individual VoIP networks for weaknesses.
Moore says Pena initially approached him to develop such tools allegedly to test his own network. Later, it became clear Pena was involved in illegal activity, but Moore continued to work for him because the money was good. He says he wanted to use the proceeds to pay medical bills for his parents who have cancer and lupus.