Abraham Williams on Google+ pointed us to an article over on the CloudFlare blog about how they dealt with a recent 65Gbps attack. The article titled “How to launch a 65Gbps DDOS attack and how to stop one” gives some high level details about how they deal with such attacks and how someone can get 65 Gbps of bandwidth to even start one. The article does a great job of explaining one method using Open Unrestricted DNS Resolver. The basic idea is that since DNS can be done with UDP packets you can easily forge the from address and cause the Open Unrestricted DNS Resolver to reply to the targeted computers or network. This is exploiting two flaws in the internet. The first that UDP is a fire and forget protocol which doesn’t require any proof of where you are coming from. The second is that Open Unrestricted DNS Resolver exist or at the least allow UDP requests. DNS can and should be required to be done over TCP which makes forging the information much harder and less reliable.
They have an article they wrote before this one that talks about and apologizes to their customers for the disruption in the first place. It’s found here and called “Post Mortem: What Yesterday’s Network Outage Looked Like” . It is a shinning example of what a company should do when an event like this happens. It is very transparent, clear and easy to understand and most of all genuine. While I know it’s great PR it’s not something I see a lot of companies like them doing.
Let us know if you have ever dealt with something like this in your job?
Do you think they took the proper response?
What do you think of the post mortem?
Update: Changed Open to Unrestricted becuase as pointed out in the comments below it seemed to imply the awesome DNS service by a similar name. They, to our knowledge, were not part of the problem.