A Broader Cautionary Tale in the Skype Outage

[Waleed Al-Shobakky; published on IslamOnline.net]

Now that the dust has settled and the uproar has faded, what can we learn from the mid-August Skype outage? (Hint: A unipolar operating system world may have unforeseen weaknesses, and Microsoft is perhaps the least to blame.)

First, what happened? Skype, the most widely used Internet phone service (free to use from PC to PC) went black on August 16, 2007. (Arak)

Now almost a household name, Skype is not merely a "chat" service for teens with plenty of time to spare. Over the past few years, Skype-in and Skype-out services that afford calls to regular land-line and mobile phones at much lower rates have already lured many businesses away from traditional telecoms.

The two-day service collapse affected thousands of businesses. Not surprisingly, the outage invited outrage.

Skype — a company eBay acquired late in 2005 for US$2.6 billion — did not respond fast. Initially they kept mum about the reasons for the service meltdown. When they spoke, they were not terribly convincing. There was talk of the impact of Windows patch prompting a massive reboot wave that the Skype software could not handle.

Always on the lookout for chances to embarrass the software giant, the anti-Microsoft camp was quick to heap yet more of their criticism on what they consider Microsoft's release-prematurely-and-patch-along-the-way model. As a result, the Skype team quickly refined their earlier statement by stressing that the Windows patch may have "triggered" the bug but was not the cause for the bug or the outage.

True, but misleading.

The Windows patch in its own right is not responsible for the Skype outage. But the desktop world's inexplicable reliance on a single operating system — Windows — is probably the real culprit.

To understand why, we need to recall how Skype works. It utilizes the Voice over Internet Protocol (VoIP) technology in which voice is transmitted as packets of data, similar to how e-mail works, with encoding and decoding happening at the two ends of the call, or the two user PCs.

The unique thing about Skype is that it largely does not have its own central servers (for routing calls and the like). Thanks to intelligent peer-to-peer network software, they harness each user PC as a node in the network. With every new user PC added, therefore, the network grows and the service (voice) quality improves.

At any point in time you log onto the Skype network, chances are you will find about seven million users online. In other words, you will be logged to a seven-million-node-strong network (possibly several times larger than Google's). The Skype founders had clearly honed their "distributed-network" model in their earlier project: the controversial Kazaa file-sharing network.

The Hypothetical Is Real

The peer-to-peer design of Skype has been, quite deservedly, touted as revolutionarily efficient. Instead of building a network that depends entirely on central servers, with costly hardware investments, you leave it to the software to connect together a network of nodes composed of the users' own PCs. And you offer the service almost free.

But a little, seemingly theoretical, question remained. How would the network function if all its users, for some reason, logged out en masse?

The hypothetical, it turned out, was real. And the answer to the question came on August 16. Users did not conspire to bring Skype down; they had to disappear from the network with the obligatory reboot after installing the Windows patch. The peer-to-peer design was put to test. And it failed (in the Skype-speak, it was only a glitch).

But before we rush to conclusions, one basic idea must be duly emphasized. This outage would not have appeared had we had a world with a variety of desktop operating system options. And it would be too simplistic to just label Microsoft as evil for this situation.

It is rather too unwise of governments, particularly in the US and Europe, to fail to see this limitation and its possible consequences. Needless to say, this does not mean that those governments should crack down on Microsoft, though this is what the EU competition court seems to be doing now. They should rather think more seriously about financing, through venture capital, and spearheading alternative desktop operating system projects.

What can be the incentive for governments to invest in operating system projects? Harnessing the collective computing power of user computers could prove crucial in affording more free and useful services, such as Skype, in both the developed and developing nations. Its potential, however, remains vulnerable as the next patch looms.

New Dimension to Old Controversy

The Skype sudden blackout indeed adds a new dimension to the long-standing controversy around Microsoft, the world's largest software company. However, this time around it is not the expensive Microsoft products versus inexpensive products from others. Nor is it closed-source versus open-source code (as in Windows operating system versus Linux). It is rather about the simple fact that a world dependent almost solely on a single computing platform is likely too vulnerable.

A technology sector with a Windows-only option is like how the world economy would be had we had, say, a Citibank-only global banking sector. There is the obvious vulnerability of not having backup or plan-B options. And there is the stifling effect a unipolar operating system world may impose on creative concepts that depend for their resilience on multi-colored networks, such as the distributed network of Skype.

Microsoft is probably the least to blame for that. And the market dynamics may not be too helpful. After all, as Microsoft grew larger — using monopolistic practices or not — it became increasingly difficult for smaller companies to compete with it. It, therefore, is incumbent on governments to step in where the market has quite failed.