Skype’s substantial period of downtime last week has been traced to overloaded servers triggering a bug in the most widespread version of the Windows Skype client, the company has reported on its blog. At the height of the problem, only a few hundred thousand users were showing up online; normally, the voice and video chat boasts in excess of 20 million online users.
Well, not Windows per se but a specific Windows client for Skype that makes up 50% of the Windows users of Skype. This crashed the computer and the way Skype is set up, resulted in bringing the whole network to a standstill.
The Skype network, and the Web as a whole, is usually configured as a small world network, with a most nodes having just a few connections and a few nodes having a lot. There are some really important advantages to these sorts networks, such as scalability. These networks can be very robust as nodes disappear or crash. Since the crashes happen randomly, most of the super-nodes are not touched and the connections can then route around any damage.
But as more nodes crash, the chances of super-nodes going down increases. If a few of these go down, then it become impossible to move information around the network at all and the network fails catastrophically. This is what happened here.
All because too many people were using a version of the client that crashed. Solving this problem could be tough. It is hard to maintain any sort of small world network when large numbers of super-nodes start failing. The increased traffic to the functional ones makes them unstable and liable to go down.
I wonder what they will do?