What happens if a domain name system (DNS) goes down? Every service — from email to HTTP — is completely unavailable, rendering websites and servers useless. So it’s no surprise that malicious actors are always looking for ways to crash a DNS server. According to Computerworld, a newly discovered flaw in popular open-source DNS software BIND9 offers just this kind of opportunity. But what’s the vector, where’s the risk and is protection even possible?
What’s in a Name?
Without a DNS, the Internet would be a very different place. Instead of easily recognizable website names, only numbers would identify pages and companies, leaving average users with the task of manually entering each and every digit to ensure proper navigation. The use of authoritative and recursive DNS servers, however, makes it easy for companies to register the name of their choice and have it automatically converted into an IP address. Open-source BIND9 is the most popular domain name software in use, making it the ideal target for cybercriminals.
As noted by The Register, this newfound BIND bug (CVE-2015-5477) allows attackers to send a specific DNS request packet that triggers a REQUIRE assertion failure. The result? A BIND exit, causing DNS servers to crash and deny access to all hosted websites. Discovered by a security researcher, the bug is so severe that a single packet can bring down multiple servers — and according to Michael McNally, lead investigator for the Internet Systems Consortium, malicious actors “have successfully reverse engineered an attack kit from what has been divulged and from analyzing the code changes.”
Versions 9, 9.1.0 and 9.10.2-P2 of BIND include the vulnerability, which has been labeled a critical fix. Some experts argue the problem lies with BIND itself rather than the efforts of researchers and attackers, claiming that the open-source software has too many features, some of which are no longer utilized by DNS servers.
Preventing DNS Server Problems
Cybercriminals have been quick to jump on the BIND bandwagon. Sucuri CTO Daniel Cid said, “We can confirm that attacks have begun.” But security teams aren’t sitting around: There’s already a patch available from Amazon, Red Hat, CentOS and Ubuntu, but deploying the patch requires admins to apply the new code and restart their DNS server. McNally noted that there are no other workarounds except patching, and he advised that other protection methods such as firewalls won’t be of any use. If companies believe they’ve been hit, it’s possible to track down evidence in a server logs, which should show the “ANY TKEY” command so long as querylog is enabled.
The BIND9 flaw comes with real risk for any DNS server. And while a patch has already been made available, the simplicity of one-packet attacks coupled with long lead times when it comes to patching means this kind of attack could linger. Like Heartbleed and similar open source vulnerabilities, there’s a long tail here. Breaking free isn’t difficult with regular patching, but worry over bound DNS servers won’t disappear anytime soon.