Despite the fact that few people know all the technical details of the bug affecting domain name servers that security researcher Dan Kaminsky reported July 8, there is no shortage of opinions on it.
As we all should know by now, the exploit discovered by Kaminsky, director of penetration testing for IOActive, takes advantage of a fundamental flaw in the DNS (Domain Name System) protocol. While Kaminsky is tight-lipped about the specific technical details of the bug, which can be exploited to redirect Internet traffic, he has said the problem has been solved by implementing port randomization.
“I’ve pinged at least a few customers, and I’m seeing very large tech firms already moving on this,” Kaminsky told eWEEK via e-mail. He said he plans to unveil details of the vulnerability at the Black Hat conference in August in Las Vegas. “Realistically, they could move pretty fast if they were under active attack. I’m trying to give people a little more time to test than that, but we’ll see what they can do with it.”
His disclosure last week was greeted with an initial round of skepticism, in part due to the fact he is keeping details of the situation close to his vest. To be sure, DNS poisoning attacks are nothing new, but if his finding is indeed on the money, it could make life that much easier for attackers.
Depending on who you talk to, keeping the details of the attack under wraps has made Kaminsky either a grandstander or a hero. Regardless, his discovery was enough for major vendors such as Microsoft and Cisco Systems to coordinate the release of a patch to address the issue. US-CERT meanwhile has included a number of workarounds and mitigations for the flaw, such as disabling recursion on any nameserver responding to DNS requests made by untrusted systems.
“It is worth pointing out that ‘split-brain DNS’ does nothing to mitigate this flaw,” Kaminsky said. “Really, if it recurses, you have to patch or you have to decommission.”
With the details unpublished, attackers may be a step behind the good guys on this one. Still, patches can theoretically be reverse-engineered. The bottom line: If you have a recursive server, you don’t have to panic, but don’t drag your feet. Always better to be safe than sorry.