Skype CIO Lars Rabbe has blamed one version of the company’s Windows client for the outage the VoIP service suffered last week. According to the company’s official blog, an IM server overload caused v.5.0.0152 to crash; since around 50-percent of users worldwide were running that version, the unexpected shutdown took 25- to 30-percent of the supernodes offline as well.
Although a majority of supernodes were still available, in-built safety measures designed to protect them should excessive traffic attempt to connect kicked in when load across the remaining servers suddenly ramped up. Windows users attempting to restart the v5.0.0152 app also added to the strain. Rabbe describes this as a feedback loop, with each crash cycle bringing down more supernodes.
Skype has already pushed out a new beta to replace the flawed Windows version, and is looking at changing its testing processes so as to avoid similar bugs making it through again. Meanwhile the paid and enterprise services will be bolstered in 2011 with a new round of investment. Paid users affected by the outage will receive free calltime.