I'm catching up on several articles and half written posts today after flying home. There's an article that caught my eye entitled VoIP Technical Details that just rankled me a bit, so I'm taking a moment to react. One of the things I'd been planning on the Realtime Community site is a series of articles on protocols. The Community is very much oriented toward business people, albeit often technical ones. I think it's important when implementing a VoIP solution, that people understand the protocols and methods being deployed. If you don't know what you're implementing or how it works, how on earth will you know if it's meeting your needs. Here's the opening that pulled me in and drove me to respond -

There is a lot of debate about the two most popular types of VoIP; SIP and H.323, each of them has its own merits. Initially H.323 was the most popular protocol, though its popularity has decreased in the “local loop” due to its poor traversal of NAT and firewalls. For this reason as domestic VoIP services have been developed, SIP has been far more widely adopted.

I guess the most troubling piece of the opening is the bogus reasoning behind decrease in H.323's popularity. H.323 declined because it's bloatware plus. The full H.323 protocol suite include just about every protocol under the sun that's ever been used for any sort of telecommunications. RTP, RTCP, T.120, H.320. H.323 isn't and never was a protocol. It's defined in the standard as "an umbrella suite of protocols."

For the most part, folks who implemented H.323 tried to implement the entire protocol suite so as to maintain as much (backward?) compatibility with legacy PSTN environments as possible. This approach brought all the bloat of these protocols, which work great in the TDM telephony workd, into IP. Bloatware. H.323 software turned out to be some of the most piggish code ever written. That's why it couldn't traverse NAT and firewalls well. It was too bloated.

To be fair, there's some accuracy in the article too. H.323 does prevail in the backbone, but that's because it was part of the backbone years ago in the subset protocols like H,22 Call Control and H.245 Media Control.

I'm sure there are plenty ot protocol tutorials around. I know there are. But over the next month or two I'll be posting a series of protocol overview articles. At the very least we'll look at SIP and H.323.for starters. I picked those because they are the staples of enterprise business VoIP systems. Proprietary protocols like MINET, CorNet-IP and SCCP can be asking for trouble, but Cisco being the 800 pound gorilla they are in IIP networking, we'll take a look at Skinny. MCGP or MegaCo pretty much died a quiet death. IAX is an interestin protocol operationally, but I'm not all that convinced it will play into the enterprise market. I'm not sure I'm convinced that open source in general will take off in the enterprise.

We'll start picking a protocol of the month to explore in depth and see what it goes. If there's one you're really interested in, now's a good time to speak up.