This in my view confirms that the people who run Skype are either incompetent, dishonest, or both. I'm leaning towards dishonest, since it's hard to imagine how anyone could be so incompetent to both (a) roll out a network to millions of nodes without simulating how the network responds to major disruptions, and (b) build a large system which cannot be promptly "rebooted" into a clean state if anything goes wrong.<p>The alternative explanation -- that Skype was the target of a deliberate attack, and they're trying to keep quiet about this -- is to my paranoid mind far easier to believe; Skype would hardly be the first multi-billion dollar company to keep quiet about an attack out of fear that shareholders and/or customers would react unfavourably.
i dont understand why people are giving skype such a bad time. they have been quite honest (technically) about what happened - a bug got exposed by some really really extreme workload. Also, whoever thinks that they should have "modeled" this has no idea how hard it is to develop distributed systems. It's a fine ideal to have, but in practice it is ridiculously hard to achieve. Especially as the code evolves. What you need is the same code base being able to run on a real TCP/IP stack and in a simulated environment. This looks easy in the abstract, but it's really tricky to do in practice.
This is going to serve as a lesson on how not to communicate a screw up to your customers in the modern age.<p>What could have simply been, "we screwed up, we're sorry" is now going to be talked about not only in terms of the event, but the poor communication strategy.<p>tsk tsk skype.