Skype Gets Lessons from Murphy

The most entertaining explanation I’ve heard about last week’s Skype outage is this posting from Rostislav Siryk in his blog:

“Skype‚Äôs outage is …[a] natural consequence of quantum physics. Because users [are] like atoms.”

In other words, it’s within the realm of possibility that all the world’s PCs will download a Microsoft update and reboot at the identical moment.

On the other hand, when was the last time you saw an object move by itself as a result of all its atoms just happening to tip the same direction?

I thought so. That’s why many are saying Skype’s explanation, issued this morning, is fishy.

Certainly, no one at Skype or Ebay is saying much. My request for a real live conversation last week was answered politely with a copy of the company’s then-current statement and a link to the Skype blog. As Skype has talked with me openly in the past, it’s thought provoking at the very least.

But this discussion begs the question. Even accepting the Microsoft-did-it explanaon, the outage is nonetheless an object lesson for the entire VoIP industry of another immutable natural law: Murphy’s.

It highlights a fundamental industry problem, says VoIP gray-beard Erik Lagerway. Providers ultimately don’t control the underlying network that delivers their service.

“I’ve been in this business 15 years and over that time VoIP has been in beta 15 years. The main reason is that the network that people are riding on is unreliable,” says Lagerway, whose VoIP pedigree includes executive roles at Shift Networks and Eyeball Networks as well as founding Vocalscape Communications and Xten Networks (now Counterpath).

Unless a provider owns the upstream broadband network, a ‘best effort’ service is all a provider can promise, according to Lagerway.

“If the upstream provider has decided they’re going to be making some changes, you’re going to be feeling those changes. If the upstream provider decides they want to filter out [other providers’ VoIP] packets or handle them with less priority than their own packets, you’re going to experience that regardless of what kind of service you have.

“If they decide they’re going to route packets to Istanbul, they can do that,” he says, adding, “The long and short of it is that the incumbents have their long arm deeply inside the network.”

Having said that, Lagerway does allow that Skype’s proprietary peer-to-peer (P2P) architecture — a closely guarded “black box” — leaves the system unnecessarily vulnerable in a way that conventional centralized services like Vonage don’t.

“My main issue with Skype is that it’s a closed system,” says Lagerway, an outspoken evangelist for the open communications standard, SIP. “Having one guy [Janus Friis] create the entire peer-to-peer architecture, it’s destined to fail — no one is smart enough.

“What’s going to happen when the next Windows update comes along? What this says is that, at any given moment, Microsoft can screw over every single Skype user. That’s a serious problem. The fact that no one even thought of this is mind-boggling.”

Lagerway points to Skype’s implementation — a self-organizing P2P network operating exclusively on users’ PCs — as untenable for providing a service to millions of users.

“To have such a dependency on so many people’s PCs, that’s pretty risky business. What happens if a whole lot of people decide to de-install?”

A better approach for a P2P network is an architecture that fails back to a centralized client-server network — the way TelTel’s P2P VoIP network operates, for example. “That’s the way SIP operates,” Lagerway explains. “It’s a peer-to-peer network but it bootstraps the operation with a client-server network.”

In the end, while no one can ever fully escape Murphy’s Law, a more open approach could have helped Skype avert this particular disaster, Lagerway says.

“If this [Skype] had been an open standards projects, you would have had much more peer review. If they had used SIP, this particular outage would have been less likely. It could have possibly been averted,” he explains. “Correcting it now is going to be costly.”

The legendary Murphy could have told Skype that, too.