Daily Payload

H.323 Continues to Lead Videoconferencing Revolution

February 19, 2010

H.323 and SIP were both created in the later 1990s to more-or-less address the same market: to enable one to place a voice or video call over an IP network. H.323 benefited from borrowing a lot of concepts from previous-generation multimedia systems, giving it the ability to easily handle voice and video conferences that might be as small as two people or as large as hundreds of people. SIP took a different direction and it seen an extremely slow road to adoption.

Those working had a naïve idea that one could create a voice and videoconferencing protocol with significantly less complexity than H.323. After nearly three years of intense work, the IETF produced the first version of SIP that had just enough functionality to enable one to place a voice call. Even then, the lack of a requirement to support at least one common voice codec meant that there was never a guarantee that any two SIP devices could communicate in practice.

Nonetheless, SIP was declared by many people around the world as the protocol that would revolutionize communications as we know it. It was declared to be a protocol that would enable all kinds of functionality that was never possible before. And what were those things? They happened to be functionality that H.323 already did and was delivering into the market. (Oh, and in case you are not aware, the first drafts of H.323 and SIP were published only a few months apart.)

And so over the next few years, H.323 continued to be deployed. It was installed by major service providers around the world to carry voice traffic. H.323 quite literally began to replace legacy PSTN network with newer IP networks. Moreover, those new H.323 networks were video-capable, which is a path the industry is still following, largely constrained only by bandwidth limitations that are only now fading away.

Meanwhile, work on SIP continued for a few more years. Finally, in 2002 a new revision of the SIP specification was published. It created a few backward-compatibility issues, but it also made some improvements to the protocol. Still, what functionality could one deliver in practice? Well, the answer was surprisingly little. The problem was that, in an effort to try to keep the protocol simple, the protocol was left very much underpowered. It had basic issues, such as dropping calls when receiving a BYE message from any rogue network device. It lacked the ability to reliably deliver messages, which caused problems such as not hearing a ringing tone when placing calls from time-to-time. There was still no common voice codec, which meant that there was still no guarantee that any two SIP devices would communicate. Further, the SIP specification did not even define how to send telephone key presses over the IP network!

Over the next several years, and with the introduction of a lot more standards, many of the basic issues with SIP were resolved. This did not come without a price, of course: complexity. In this regard, SIP is like a wolf in sheep’s clothing: it appears simple on the surface, but is in fact a very complicated protocol. After all, SIP aims to enable voice and videoconferencing functionality and any such system inherently has a certain amount of complexity. It takes about the same amount of complexity to do the same thing two different ways.

All the while, H.323 continued to be deployed. It has also matured over the years and H.323 version 7 was standardized by the ITU in late 2009. There are open source products on the market and, with the rise in interest in videoconferencing, H.323’s strength as a protocol built on the foundation of successful videoconferencing technology makes it very appealing. H.323 is a modern IP-based communication system that allows one to place calls to people using traditional phone numbers or URLs like user@example.com. It fully supports inter-domain federation, meaning that users all over the world can use H.323 in order to communicate and hold video conferences over the Internet.

It has long-been said that H.323 is dead, yet there is no clear indication of that. Some cite network technologies like IMS as evidence that H.323 is being replaced by SIP, yet many service providers do not deploy IMS in practice. Rather, many service providers use SIP just for basic SIP voice calls in order to mimic the PSTN over IP. H.323 can and does do the same thing, so the very decision to employ SIP has largely been one based on hype, not features or functionality, and certainly not time-to-market. And why would one want to build a complex IMS-based network when any such network that runs IP may very well see its users choose other technologies, like Skype?

Meanwhile, most enterprises around the world that operate video equipment are looking at H.323 and trying to further evolve the installed base. There is a renewed interest in interconnecting videoconferencing systems over the Internet, with H.323 as the protocol of choice. It is, after all, a protocol designed for videoconferencing, designed to operate over IP, and had benefited from a number of years of maturity and stability.

It has been interesting to see the continued claims of H.323’s demise, yet the future of H.323 looks more solid today than it did 10 years ago. There will be those who continue to speak of SIP’s strengths and advantages, but feature parity with H.323 and the overwhelming complexity of IMS is a problem for SIP. Moreover, by the time SIP does gain feature parity, would it not be time to look forward at the next major evolution in technology?