I was at a gig this evening when an interesting tweet showed up in my timeline. It said “Major internet outages across America” and linked to a website displaying stats on global internet traffic. The stats for North America showed something odd. Several nodes were not responding, response times were up significantly, packet loss was up and consequently traffic was dipping big time. I thought at first that it might be due to hurricane Sandy but when I looked at the locations they were all over the USA. Weird…
I tweeted a few people in the US to see if they had any idea but nothing concrete came back. Some suggestions were made that the traffic drop might be due to lunchtime but that didn’t make sense to me. Especially as lunchtimes across the US are obviously spread out and also because lunchtimes would not cause excessive packet loss. So I decided to have a look at the stats for Europe. To my surprise the same thing was going on there:
Next were the global stats which showed the same story:
So there has been a substantial increase in response time and packet loss resulting in a big dip in internet traffic, across the globe AT THE SAME TIME…
That was a serious Whiskey Tango Foxtrot situation. So I had a look at the global stats for the last 30 days. These stats indeed confirmed what I described above but it also showed a few other interesting facts;
- An even bigger blip had occurred around October 8-9.
- There has been a steady increase in response times & packet loss since October 8th.
Now there might be perfectly harmless explanation for this but we are looking at a homogenous network across the global with components managed and owned by a mix of private and public sector organisations. To see a global trend in performance across the whole network for a prolonged time is something that raises questions. So I decided to run a quick search for other reports on this and stumbled across an article by ABC news. The article basically reported in a bit more detail what I outlined above. It reported that Youtube, Amazon, Google’s App Engine, Tumblr and other sites had been affected. However most interesting was that there was no explanation offered for what might have caused this degradation.
I will keep digging..