AT&T said the company was hit with a distributed denial of service attack (DDoS) Aug. 15 that disrupted Internet service for some of its customers.
The attack, which reportedly lasted several hours, affected domain name system servers belonging to the company.
“Due to a distributed denial of service attack attempting to flood our Domain Name System servers in two locations, some AT&T business customers experienced intermittent disruptions in service on Wednesday,” a spokesperson told eWEEK. “Our network and security teams quickly worked to mitigate the impact and service is currently running normally.”
“We apologize for any inconvenience to our customers,” the spokesperson added.
The company did not offer information about whether it is boosting its DDoS defenses in light of the attack.
“DDoS attack mitigation is a very complex and dynamic cat and mouse problem,” said Jeff Lyon, president of security company Black Lotus. “Those who design attack vectors for the specific purpose of circumventing defense measures are equally, if not more, intelligent than the security engineers implementing solutions to the problem.”
Several years ago, DNS reflection attacks and TCP SYN floods would have been the most difficult attack vectors, he explained. The major challenge today, however, is with complex attacks at the application layer (Layer 7) that deplete resources of the server itself by mimicking real users in a way that is very difficult to detect. In recent months, he added, attackers have begun targeting Websites that rely on SSL by exploiting a Web server’s ability to handle large amounts of HTTPS sessions.
“Increasingly, we are seeing that DDoS attacks against high-profile targets are intelligent, determined and persistent,” said Neil Roiter, director of research at Corero Network Security.
“It is therefore essential the DDoS response plan defines when and how additional mitigation resources are engaged and surveillance tightened,” he said.
A brute-force or flooding DDoS attack is relatively easy to identify but it requires high performance and sophisticated real-time analysis to recognize and block attack traffic while simultaneously allowing legitimate traffic to pass, Roiter said. Detecting application layer attacks requires a thorough understanding of the typical behaviors and actions of bonafide customers or employees accessing the applications being protected.
“In much the same way that credit card fraud detection may be automated, on-premises DDoS defense systems establish legitimate usage profiles in order to identify suspicious traffic and respond accordingly,” he said.
In addition, Roiter said organizations should make it a priority to protect their DNS servers.
“The Internet Domain Name System (DNS) is a distributed naming system that enables us to access the Internet by using recognizable and easy to remember names such as www.google.com rather than numeric IP addresses (e.g. 192.168.0.1) on which network infrastructure relies to route messages from one computer to another,” he said.
“Since DNS is distributed, many organizations use and maintain their own DNS servers to make their systems visible on the Internet. These servers are often targeted by DDoS attacks. If the attacker can disrupt DNS operations, all of the victims’ services may disappear from the Internet, causing the desired Denial of Service effect.”