Sub7 vid Trojan can launch distributed attacks
A terrible beauty is born
Last week we expressed doubts about a report from security outfit NETSEC, claiming that they had found a new Trojan capable of launching DDoS attacks.
Their "new" Trojan turned out to be Sub7, a remote administration package which had been around for years, and which we deemed an improbable candidate for DDoS.
After discovering first that NETSEC was wrong about the novelty of their discovery, and after assessing the relatively low threat Sub7 posed in the DDoS arena, we drew the natural conclusion that the company was yanking the media's chain for attention.
Subsequent e-mail correspondence between The Register and NETSEC executives further persuaded us that the company did not have its facts straight, and was scrambling for after-the-fact validation of its original claims.
Now we learn that NETSEC was on the right track after all, and if they had simply waited until they had a firm handle on their find instead of disgorging inaccurate data through the media in their rush to get attention, they might have spared themselves a significant PR cock-up, and won some serious props in the security community.
As it turns out, the most recent build of Sub7 contains an undocumented feature which can indeed be used to ping the living hell out of Web servers, from numerous infected clients simultaneously, according to research just completed by security outfit iDefense.
Sub7 has long used an IRC feature which logs the infected servers into an IRC channel of the operator's choosing, to notify the operator of which victims are on line. At that point the operator can log on to a victim's computer using the Sub7 client, and go about whatever remote administration tasks he had in mind.
A later feature configured the IRC bots to listen for commands entered in the IRC channel, which would be executed simultaneously by all the victims logged into it.
Now for the interesting bit: Although the Sub7 crew has decided not to document it, IRC bots in the new build will listen for and respond to ping and mping commands, iDefense Chief Scientist Sammy Migues told The Register.
So, if one has managed to infect, say, a thousand victims, and could reasonably expect perhaps 250 of them to be on line together at any given time, one could run an mping command through all of them simultaneously.
An attacker can choose a target IP; command the 250 victim machines to send say, one million packets of 64K each; and, voila, an instant, and distributed, ping flood.
The complete iDefense report is posted here in PDF format, for those who wish to acquaint themselves with all the gruesome technical details. ®
Sponsored: Global DDoS threat landscape report