Skype provided more details about why its voice over IP service has been affected yesterday and today and it highlights the very important role of supernodes which act like the middlemen to connect Skype users.
The company's Peer to Peer telephony technology should have, in theory, allowed it to function without disruption even if computers go down. Clearly, the service did not buckle under pressure on the build-up to Christmas but then, it remains a mystery why so many supernodes were taken down almost simultaneously.
Supernodes represent a tiny fraction of hundreds of thousands of computers that form the Skype network; the company says of supernodes that "a small percentage of our users will hold a record reflecting the online presence of other users. When one user holds a record concerning the presence of other users, the former is called a supernode or directory node."
It is not clear how Skype chooses which computers should become supernodes and whether supernodes are constantly change (although we doubt it) but users are certainly not aware of the fact that their computers may have become a node directory.
Skype says that the issue arose from a problem affecting "some versions of Skype" without giving much details. What we do know is that since Skype 3.0, there is an explicit switch in the client that allows users to disable the supernode functionality.
If hackers can somehow determine the few hundreds (or thousands) computers that make up the supernodes, then they could potentially either block their internet connection or activate that kill switch (via a Windows Registry).
Skype said that its engineers were busy creating new "mega-supernodes" as fast as possible, which brings up the question of "why didn't they do that before". Hopefully, Skype will deliver a complete post-mortem of what caused the downtime.
In the meantime, the number of people on Skype is growing at the rate of around 100,000 every five minutes with roughly 6.4 million users accounted for by Skype.