Slashdot Mirror
Michael Robertson Talks VoIP With Voxilla
Vick writes "Two 'Bad Boys' of internet audio, MP3.com's Michael Robertson and Kazaa's Niklas Zenstrom, are done taking on the recording industry. Now their big fight is with the telephone companies and, apparently, one another. In one corner is Zenstrom's Skype, a software-only VoIP product that uses its own protocol and is banking on the huge popularity of Kazaa for its success. In the other corner is Robertson's SIPPhone.com, trying to simplify VoIP, and using the standard SIP protocol, to try to bring internet telephony to the masses. In this Voxilla.com interview, Robertson talks about the future of VoIP and minces no words in explaining why Zenstrom and the Kazaa boys have got it all wrong." (Last month, we posted about Skype.)
SIP is a P2P standard. It pretty much follows the IETF party line of "Lobotomized Core, Smart Edge" in design -- all the intelligence in SIP call handling is supposed to be at the end points, handled by the peers. It tries very hard to make the network itself brain dead.
It gets fuzzy around the REGISTER/location servers -- until every ISP has a SIP REGISTER/location server per domain name and/or hostname, SIP users are going to need some place to go where they can look up and find each others' network information/location data so they can place calls to one another. So at the moment, that's tending to push a centralized REGISTER/location server from the current VoIP providers.
There are a host of issues with running VoIP in a P2P model. NAT traversal being just one of them. Another is the authentication/authorization problem -- SIP was at once point locked in endless arguments about IPSec vs. TLS, certificate management among all the end points, key management, etc.
... not so open your brain falls out.
The flip side is that sometimes a standard can be *too* open, too easily built upon, and it fragments because no one can agree on exactly what it is any more.
User_A has a SIP phone that supports session-timer, OPTIONS, REFER, and NOTIFY. User_B has a SIP phone that supports rel-100, PRACK, SUBSCRIBE, and NOTIFY. User_C has a SIP phone that supports MESSAGE, session-timer, and OPTIONS. All of these are SIP phones, and you could probably make a phone call between all of them, but will any/all more advanced services work between them? Hard to say, but each manufacturer behind the phones can call it a fully compliant SIP 2.0 phone.
If the standard is toooo open, everyone will have to be checking the labels to see which extensions each phone has to make sure it interops with another one... or they will all just default to buying the well known brand name/proprietary implementation, if only because they know it will work with everyone else who bought the same brand name.
The obvious attraction of VoIP is not enough on itself to make it succesful, rather it will need a big push in order to get going. All I have seen so far is that it has barely advanced beyond the simple voice chatting features of an IM client such as ICQ.
The interesting thing about SIP is that it is a generic protocol for starting and running conversations. It's not limited to one medium like say Jabber is limited to text IM only. You can use SIP for IM, VoIP, videoconferencing, file transfer, shared whiteboards, whatever you want. It's pretty cool. And it has loads of real-world vendors behind it. Forget about dodgy startups like Kazaa, I'm talking Nortel, Alcatel, Microsoft, et al. That's important because these are companies that ship real products (i.e. phone on your desk, the phone switch in the basement of your office, etc etc). They can simply roll SIP in and migrate customers very smoothly to it. The analogy to MP3 doesn't really hold, because the real strength of SIP from a consumer perspective is that it will be transparently embedded in everyday items - most users will never even hear about it.