Research Experiment Disrupts Internet

CommanderFrank

Cat Can't Scratch It
Joined
May 9, 2000
Messages
75,399
If your internet connection was a little wonky yesterday, you can blame it on an experiment run by Duke University and RIPE NCC in Europe. It only lasted about half an hour, but shows the vulnerability and fragile nature of the internet as a whole.

"Over 3,500 prefixes (announced blocks of IP addresses) became unstable at the exact moment this 'experiment' started," wrote Earl Zmijewski, a general manager with Internet security firm Renesys. "Not surprisingly, they were located all over the world: 832 in the US, 336 in Russia, 277 in Argentina, 256 in Romania and so forth. We saw over 60 countries impacted."
 
Yeah, they ran their experiment at the exact same time I ran my global bandwidth outage test, which transmitted over 30 terabytes of data a second. Once I started my experiment I noticed issues with the net infrastructure, their test messed with mine!
 
Couldn't there be legal issues with this? If they knew their "experiment" would result in world wide disruption,it could be interpreted by some as a criminal act.
 
Taking down the internet would be the craziest shit to happen in recent history imo.
 
Interesting. I had a lot of problems getting onto several websites at work that morning.
 
The Matrix has been switched on...You know what comes next.
 
Something happened to Verizon at like 1 am last night Eastern Time.... was this part of it?
 
It was tagged as experimental. So I don't think they should be liable. Good thing they are not located in the US though.
 
It was tagged as experimental. So I don't think they should be liable. Good thing they are not located in the US though.

If it disrupted telecommunications for credit card companies, banks, or worldwide stock exchanges someone is definitely going to sue and win and it won't be a small amount.
 
This is the main reason the internet needs the person who is in charge to bring in substrates to guard and cover each individual network infrastructure allowing bots through and blocking money making schemes.
 
It was tagged as experimental. So I don't think they should be liable. Good thing they are not located in the US though.
Duke sucks and is most definitely in the US.

Nevertheless, according to the article, the experiment data was up to standards but exposed problems with certain router protocols. As such, it is the router vendors who should be nailed to the wall.
 
^ I should say "exposed problems with the implementation of certain router protocols"
 
Regardless if their experiment was up to standards, why would you run experimental anything in a production environment? Seems to me that BGP route testing should be done in a test environment so you can see first hand the results, rather than hearing about issues you caused from ISP's downstream.

I would say that they exhibited poor judgement and planning for this exercise.
 
Regardless if their experiment was up to standards, why would you run experimental anything in a production environment? Seems to me that BGP route testing should be done in a test environment so you can see first hand the results, rather than hearing about issues you caused from ISP's downstream.

I would say that they exhibited poor judgement and planning for this exercise.
Experimental test beds that emulate the internet are available. Unfortunately, these would likely not include misconfigurations in the 1% of internet routers that caused the problem. (In fact, I would go further and say these test beds probably were employed in a Proof of Concept first.)

An in-lab research project may very well determine a more efficient way to, say, propagate routing information. How comfortable would you be rolling that out Internet-wide without a real-world experiment first, hmm?
 
Back
Top