The great open source VoIP debate

Two weeks ago at the Internet Telephony Conference & Expo in Miami, I had the pleasure of moderating the open source and telephony roundtable discussion with some of the key players in the business. The panel included Mark Spencer, creator of Asterisk and CEO of Digium; Bill Rich, CEO of Pingtel; and Alan Hawrylyshen, CTO of Jasomi Networks.A debate, which was sometimes heated yet always mutually respectful, illustrated philosophical and methodological differences among the three men but left no doubt that each had an important contribution to make in open source and VoIP. The panel fell into two camps on open source VoIP--Asterisk and SIPFoundry.

Asterisk is a complete telephony package that provides an analog PSTN and IP-based telephony solution. Asterisk is probably the most versatile PBX in the telephony market. It offers support for just about every kind of VoIP protocol that you can imagine--including IAX, SIP, MGCP, H.323, Cisco SCCP, and the recently added Nortel UNISTIM. It has support for virtually every VoIP Codec, too. (However,one of the most desirable Codecs, G.729, must be licensed due to its patent-encumbered royalty status. Therefore,many in Asterisk and the open source community preferSpeex because of its royalty-free status.) If all that wasn't enough, Asterisk boasts an extremely wide array of support for analog telephony, which is still important if you want to talk with the rest of the world and their primitive analog phones. This versatility is probably the biggest appeal of Asterisk.

In one of the lighter moments during the discussion, Jasomi Networks' Hawrylyshen recounted first meeting Mark Spencertwo years ago at a conference. At the beginning of that conference, Asterisk had nothing for SIP, but at the end of that conference a few days later, Spencer had added SIP capability to Asterisk. The other distinguishing aspect of Asterisk is that it has its own protocol called IAX, which unifies VoIP signaling and payload into a single UDP stream that can easily traverse firewalls and NAT devices and includes built-in multiplexing capabilities that can essentially act as a bandwidth multiplier when transmitting multiple voice streams. Unlike SIP, which requires separate signaling and data ports and lacks native multiplexing capabilities, IAX doesn't require any fancy devices and protocols to solve the firewall and NAT traversal issues that plague SIP communications. The problem with IAX is that it currently doesn't have the sheer clout that SIP has in the IP telephony industry and there are currently few soft or hard IP phone clients that support it. Nevertheless, the success of Asterisk does not hinge on IAX becauseof its sheer flexibility to support nearly every protocol under the sun. Since IAX is already useful and very efficient in a number of ways--such as PBX-to-PBX communications between branch offices or trunking to a PSTN gateway service provider--the race between IAX and SIP may be more of a marathon than a sprint and there is always enough room for more than one protocol.

Pingtel can be thought of as (in Bill Rich's words) "the Red Hat" of the IP PBX market by providing a polished and supported open source telephony solution. Unlike Red Hat, which simply markets and packages Linux, Pingtel is also one the largest contributors of code to the SIPFoundry project.Pingtel really puts their money where their mouth is in terms of open source. Since Pingtel is heavily involved in SIPFoundry and the IETF SIP standards, which is quickly becoming the de facto standard in communications, Pingtel is in a position of great influence in the IP telecom world. When I noted that all the bigplayers in the PBX market--like Nortel, Avaya, and Cisco--have their own proprietary protocols for VoIP, Bill emphasized the dominance of SIP with the fact that it has been adopted inthe roadmaps of every major PBX maker. The same can't be said about any other protocol. Users now have a choice to go to SIPFoundry and download the free open source IP PBX solution that they put together themselves, or they can go with Pingtel's commercial solution. Where Pingtel's commercial solution shines is that it is probably one of the more polished in terms of manageability and ease of use. Pingtel goes out ofits way to coddle users who are not comfortable with setting up Linux and compiling open source code. In addition to the hand holding that Pingtel customers get with their support, they also get a complete bootable CD that installs Linux and all the other necessary IP telephony components,allowing a mere mortal to turn an ordinary server in to a full-blown IP PBX. Although Pingtel and SIPFoundry are focused purely on SIP and IP telephony with no code to support analog telephony interfaces on their IP PBX, Pingtel does extendits management interface to support most of the popular Media Gateways such as Cisco routers with telephony interfaces.

Jasomi is a company that was a founding contributor to one of the SIPFoundry projects and continues to remain active in the SIP Foundry community. With products that allow SIP to traverse firewalls and NAT boundaries, the companycaters mostly to service providers and large enterprises.Hawrylyshen, who represented Jasomi, not only made agreat case for his company, but also illustrated how it was possible for a business to leverage open source without giving away the family jewels. He explained how it is possible for a company to keep enough intellectual property for its very survival while navigating the various types of open source licenses and their legal obligations.His eloquent presentationgave me a much better understanding of open source in general.

Near the end of our panel session, I turned the discussion to debate the merits of IAX versus SIP and things got interesting very quickly.Astrerisk's Mark Spencer pointed out that SIP was overly complex while Hawrylyshen defended the merits of SIP against IAX as more of a general purpose protocol that is more than just about making phone calls. Hawrylyshen argued that the general purpose protocol will always be less optimized than a more narrowly focused protocol. He suggested that SIP was so broad that it could even adopt IAX within it. This, of course, prompted immediate interest in the audience;one audience member asked if anyone could write an IAX module and contribute it to the SIPFoundry project. Hawrylyshen responded that there was no reason why not. As excited as the audience seemed about this theoretical unification, none of the panelists seemed too eager about its prospects. At this point, I was already running 10 minutes late and I had to end the great debate. In the end, I walked away with a greater understanding of the state of open source and VoIP and respect for each of the panelists and their companies. What are your impressions? Talk back here.