Why Skype won't drag eBay down

In his Corante.com post The Fall of eBay, longtime friend and ZDNet Open Source blogger Dana Blankenhorn seems to view Skype as an albatross on eBay's stock price and even long-term viability.

As a service, eBay has problems, but Skype isn't one of them. Let me explain by reviewing what I consider Dana's key point.

Dana quotes Info-Tech research analyst Ross Armstrong saying - and I am paraphrasing - that Skype is so insecure, corporations shouldn't get within about seven parsecs of it (parsec=3.2 light years). Insecure because it is peer to peer, and besides that, you, the corporate IT, would have to write all sorts of rules and regs for it.

Well, to that, Ross is somewhat off the mark. Skype's business use currently accounts for about 25 percent of all its traffic. Have you heard of even one instance of Skype being hacked, of a security violation that according to Ross' scenario, ought to be rampant?

No, and that's because even without firewalls, non-firewalled clients and clients on publicly routable IP addresses are able to help NAT’ed nodes to communicate by routing calls. Not only does the technology allow two clients that otherwise would not be able to communicate to speak with each other, but the encrypted nature of these calls from end to end inform the use of proxies that limit the security or privacy risk.

There's another key point that Ross may be missing. So much of enterprise connectivity doesn't get carried over the public Internet, but via secure Virtual Private Networks and even plain vanilla T-1 lines. It is more than possible to port Skype calls over these pipes.

As to the motives behind Ross Armstrong's arguments, one poster to Dana's thread has it on the mark.

Jesse Kopelman's main points seem to be that Skype is valuable to budget-conscious enterprises, and all that "shut Skype off now" is a tone to be expected from analysts who will sell more reports if they can posit scary scenarios.

The people who want corporations to turn off Skype are the same people who wanted corporations to turn off WiFi, not provide Internet access to general employees, not move from dumb terminals to PCs, etc. Then there are the analysts whose business it is to sell reports and the best way to sell them is to have some new threat that people need to be made aware of. Fear is a powerful motivator. Nothing comes for free. Any new technology poses security challenges and the question is whether or not they are outweighed by productivity gains. I work with someone who was able to go from a $200/month to $100/month cell plan because of Skype and with the savings he added unlimited EVDO and further increased his productivity.

And as to access policies, many third-party Skype add-on utilities push the ball up even further in this area.

So the problem is not security. If there is a problem for Skype in the enterprise, it is a lack of robust QoS of the type offered by traditional enterprise-strength VoIP solutions such as by Avaya and Cisco.

Dana's other key point seems to be that since the regulatory trends and political winds seem to favor monopolistic broadband providers - and providers such as SBC (which he doesn't specifically cite) who want to block Skype access will probably be able to get away with it, then eBay cannot count on a robust revenue stream from Skype.

Actually, what SBC wants to do with on-the-grow initiatives such as Project Lightspeed is to charge content suppliers for bandwidth assurances and quality of services. Right now, the best thinking on this issue is that of the aggregate minimum rate of 25mbps that will be offered to each household, these enhanced services- including IPTV- will gobble up about 5 or 6 mbps.

That leaves plenty of room for Skype. Let me explain. VoIP bandwidth consumption depends on the codec that is used. Plus, every channel is not used all the time, and most conversations include moments of silence. In a Skype conversation, that means packets go down as voice does.

As to codecs, the G.711 codec is among the most robust today. SunRocket and SIPphone (Gizmo Project) use it. Typical NEB (Nominal Ethernet Bandwidth) for this codec is about 87.2 kbps. It would be just a tiny percentage of the 19 mbps left over in Lightspeed after that guaranteed 6 mbps was used. To be exact, 0.46%.

But Skype doesn't even use G.711. It uses a codec called iLBC (Internet Low Bitrate Codec). iLBC's NEB is about 27.7 kbps. So what percentage of 19 mbps is 27.7 kbps? About 0.15%, or about 1/550th of what's left after Lightspeed's prioritized services get their piece. I don't think SBC would mind...

There's also a fundamental difference between P2P-type Skype-to-Skype calling, and the type of packetized calling exemplified by Skype's PC to PSTN offerings such as SkypeOut.