The two-day outage last week of Skype peer-to-peer VoIP services was triggered when users of the software installed Microsoft patches and rebooted their computers, which also serve as nodes in Skype’s network.
The incident demonstrates a reliability weakness in the company’s network architecture and indicates that Skype’s technology might not be well suited for corporate use, especially if phone calls are critical to doing business.
According to Skype, its network -- which relies on customers’ PCs that are networked to help handle call setup -- became disrupted. “The disruption was triggered by a massive restart of our users’ computers across the globe within a very short timeframe as they re-booted after receiving a routine set of patches through Windows Update,” according to Skype spokesman Villa Arak.
In addition to tying up user machines, the restarts were clustered and triggered a flood of log-in requests in a short timeframe, and these requests overwhelmed the weakened network, Arak says.
The company says its network is designed to fix itself based on a network resource allocation algorithm that distributes traffic among available machines. The combination of machines being tied up by Windows Update and a surge of log-ins found a weakness in the allocation algorithm that prevented it from responding fast enough to head off an outage.
Skype says it was not the victim of malicious behavior, and that the security of its customers’ machines was never at risk.
Skype is interesting to some businesses because its use of the Internet rather than phone networks can save money, and Skype supports business features such as teleconferencing and videoconferencing. Skype has even given its customers limited access to provisioning its software to limit the features allowed on corporate networks. Uncontrolled Skype use in a corporation can lead to dramatic bandwidth drains.
Read more about voip & convergence in Network World's VoIP & Convergence section.