Ensure Ease of Recovery with Asigra’s Agentless Software

Last month's attack on at least six of the net's root servers was formidable, but thanks to the implementation of a technology designed to protect the infrastructure, only two were affected, according to a factsheet issued today by ICANN.

The DDoS attack flooded the servers with a stunning amount of data, as much as 1 Gbps at points, according to the oversight group. But damage was relatively contained thanks to new load-balancing technology called Anycast, which was installed on all the servers that came under fire, except for the two that sustained damage.

That allows the servers - which translate domain names such as theregister.com into IP numbers such as 72.3.246.59 - to distribute large volumes of data evenly among many machines (a many-hands-make-light-work approach, if you will). Anycast also safeguards against damage that could be caused by natural disasters by geographically dispersing servers.

The attack was, in fact, two forays, one that started at 4AM California time on Feb. 6 and lasted for two-and-a-half hours, and the other, a five-hour assault that started several hours after the first one stopped. That torrent of packets originated from hundreds of zombies, so it's impossible to know where in the world the attackers were located. Educated guesses lead engineers to believe they came from the Asia Pacific, possibly in Korea.

Of the two servers that sustained damage, one was the g-root, which is operated by US Department of Defense and is physically located in Ohio, and the other was the l-root, run by ICANN and based in California. ICANN attributed the damage they sustained to the absence of Anycast.

In addition to those two root servers, three others also are not using the technology. Engineers have held off installing it universally because some worried there could be risks associated with making different servers appear to be coming from the same place. "And so just a few root servers tried the system first, tested it thoroughly and ironed out any bugs before the next set moved over," the report explained. (Note: In previous versions of this story, we misread the report to say the delay was based on root servers using different OSes. This is not the case.)

After last month's episode, it's likely the remaining servers - which in addition to roots G and L also include roots D, E and H - will adopt the technology soon. ®

Requirements Checklist for Choosing a Cloud Backup and Recovery Service Provider