http://arstechnica.com/news.ars/post/20080827-inherent-security-flaw-poses-risk-to-internet-users.html
8/29/2008 1:33:53 PM
The responsibility to stop this from happening rests on the shoulders of providers.There are plenty of tools available within bgp as it currently exists which allow a provider to block its customers from advertising prefixes it does not own.If every provider conformed to these standards, then this vulnerability would not exist.Even if some of the major ISPs were to create some manageable borders in their networks where they could isolate the potential reach of such a hack, the ability to do this successfully would be severely hampered.
8/29/2008 2:14:51 PM
Hey, BGP is pretty good for something that was drawn up on a napkin in a restaurant
8/29/2008 3:12:38 PM
wait what
8/29/2008 3:25:25 PM
http://www.youtube.com/watch?v=HAOVNYSnL7kDr. Rekhter gave that talk while I was at the Goog. Very awesome, and definitely worth a watch! He's quite a humble guy, considering the impact that he's had on the networking world.I hadn't read anything about this 'flaw', but just figured I'd throw that in given the two previous comments.
8/29/2008 6:31:09 PM
Another article that provides a bit more depth:http://blog.wired.com/27bstroke6/2008/08/revealed-the-in.htmlrobster hit the nail on the head. I don't agree with labeling this as a "fundamental design error" in BGP. IMHO I don't see this as anything new, just another mitm with a touch of bgp traffic engineering.
8/29/2008 6:53:25 PM
the problem is that this isn't really seen as a huge security risk so not all providers allocate the resources to fixing the problem. and some providers are small and incompetent and/or dont have the skills to manage their networks properly. for example, youtube accidental blockage earlier this year when a government isp null-routed youtube on their network.
8/29/2008 10:39:48 PM