Threat From DNS Bug Isn't Over, Experts Say

The Domain Name Server design flaw that threatened the entire Internet earlier this year has mostly been patched, but the threat is far from over, experts say.

The DNS flaw, which was discovered by IOActive researcher Dan Kaminsky in the first half of this year and resulted in the largest simultaneous security software patch in Internet history in July, was fully disclosed last month. The flaw makes it possible for attackers to exploit the recursive nature of DNS server queries to “hijack” TCP/IP sessions and potentially redirect large segments of Internet traffic to unintended destinations. (See Vendors Issue Massive Simultaneous Patch for Common Internet Flaw, Details, Exploits of Web-Wide DNS Vulnerability Revealed, and Kaminsky: DNS Vulnerability Will Affect Email, Internal Systems, Too.)

The simultaneous patch, which added a port randomization factor to the transaction ID used to authenticate Internet sessions, reduces the likelihood of session hijacking by many orders of magnitude. And most DNS servers have deployed the patch -- the number of servers vulnerable to the DNS attack has been reduced from more than 85 percent to fewer than 30 percent, Kaminsky said after conducting some vulnerability testing last week.

But the patch doesn’t fix the DNS flaw -- it only makes it more difficult for attackers to exploit it, Kaminsky and other experts are quick to say. While hackers ramp up their DNS exploits, DNS server makers and operators likely face another round of patching that will occur before the end of the year.

“We will probably see the next round of patches in the next one to three months,” Kaminsky said in an interview last week. This round probably will not be done simultaneously by all DNS server vendors, because it involves more product- and vendor-specific patching than port randomization, which could be globally applied to all DNS server environments. The next generation of patches will likely include some capability to put DNS servers into a defensive posture when they detect an onslaught of traffic attempting to exploit the flaw, he said.

And that type of traffic has already been spotted on the Internet. Paul Wood, a senior security analyst at MessageLabs, said his company has seen a 52 percent increase in traffic that appears to be seeking out recursive DNS servers that might still be vulnerable to attack. “This attack requires a large amount of traffic,” which makes it easier to spot, he said.

Attacks that exploit the design flaw have been spotted in the wild as well as on Metasploit, although most of them are designed to attack systems that have not yet been patched. Experts are concerned that the next generation of attacks will attempt to compromise the patched systems, which would be much harder but is certainly possible, they say.

“What we’ve got out there so far are truly Band-Aids,” says Alan Shimel, chief strategy officer at StillSecure, which has been monitoring the DNS flaw since it came to light. “We’re putting our finger in the dike. There are questions on how to move the solution to the firewall level. There’s no way to ensure that port randomization is being done. We need the next generation solution. We need a new DNS.”

Kaminsky agreed. Even after the next round of patches, the flaw in DNS will still be there, and the most frequently cited longer-term solution -- DNSSEC -- will require broad adoption before it can be effective, he noted. “It requires authoritative servers to opt in, and that will take time to happen.”

In the meantime, attacks will be difficult to spot, especially on unpatched DNS servers, Kaminsky observed. “It sucks to scan for it. The attacker can grab ten seconds worth of traffic and then wipe the cache. The attacker basically controls how much log data there is.”

Kaminsky says he is most concerned over the potential of an attacker to corrupt an email name server, essentially allowing him to redirect large amounts of email traffic. Companies should be sure patch the name servers that support their email servers, he said.

MessageLabs’s Wood agreed, adding that users should take steps to encrypt sensitive email traffic, so that even if messages are successfully diverted, they will be difficult to decode. “And you should keep hold of the keys, and be sure reset them when employees leave,” he said.

“This is a long way from over,” Shimel says. “We’ve built a moat around the castle, but there are a lot more issues here, and I’m not aware of any security company that’s directly addressing them all.”

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.