Until Skype Froze Over

Thursday, August 23, 2007

Skype’s popular internet telephone service went down on August 16 and was unavailable for between two and three days.  Skype relies on peer-to-peer technology, where each client is also a server, reducing the load on centralised servers.  Peer-to-peer applications are meant to be resilient: if one peer fails, another can take its place.  Skype’s extreme outage shows that this is not always the case.  So what happened?  Skype says the problem was triggered by a Microsoft patch, delivered by Windows Update, which caused an automatic reboot of many PCs.  “The high number of restarts affected Skype’s network resources.  This caused a flood of log-in requests, which, combined with the lack of peer-to-peer network resources, prompted a chain reaction,” says Skype’s Villu Arak.  According to Arak, the system would normally have recovered quickly, but on this occasion “a previously unseen bug” caused the network to fail.  Some observers have speculated that Skype’s entire distributed database became unsynchronised and had to be rebuilt, accounting for the long delay.

The incident prompts a number of awkward questions.  The first is whether it is sensible for a company to make critical services dependent on its customers’ machines – over which it has no control.  Arguably, the Skype incident was merely an unfortunate combination of lax software quality exposed by an external trigger rather than a problem with the peer-to-peer concept itself, but it still proves that distributing an application over millions of machines does not guarantee resilience.  Suddenly the business benefits of peer-to-peer seem more doubtful…..Another issue is that probably only a few Skype customers fully realised what they agreed to when they clicked OK to the licence agreement.  In Skype’s case, any machine may become a supernode, which is a server responsible for hundreds of other clients.  Similar considerations apply to the BBC’s iPlayer application for TV over broadband and Sky’s similar Anytime service, both of which use a peer-to-peer application called Kontiki.  When you download a broadcast, you are really downloading it from another user’s PC, as well as allowing others to upload it from you.  Both Sky and the BBC are rather quiet about this aspect of the service, which could be a significant problem for anyone paying directly for data transferred, or simply concerned about others using their PC as a server…..Skype’s troubles are good news for some.  Blogger Om Malik reports that rival VoIP provider Gizmo Project saw a four-fold increase in sales during the outage.  The Gizmo Project uses the standard Session Initiation Protocol (SIP), in contrast to Skype’s proprietary system.  This means that unlike Skype, the Gizmo Project software can place calls through other SIP-based providers.  At times of stress, it can pay to have multiple options.

Reference : http://www.guardian.co.uk/technology/2007/aug/23

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: