free H.323 logo
free h.323
SIP

Credit
The following letter, addressed to the OpenH323 mailing list, by Greg Herlein, inspired the creation of this page:

Date: Wed, 15 Sep 1999 07:58:58 -0700 (PDT)
From: Greg Herlein
To: OpenH323 mailing list
Subject: Re: [openh323] Thoughts on H.323 and SIP Telephony

I have to toss in a few thoughts myself:

H.323 is a huge family of protocols that covers the gamut of things one would need to do to support IP telephony. The core documentation is something on the order of 1400 pages of highly technical stuff. H.323 includes gobs of stuff on data sharing, conferencing, video, etc.

SIP has a fairly limited set of cpabilities and is pretty much focussed on IP telephony. I think the core protocols (counting SDP which is used in SIP) is about 250 pages of moderately technical material.

H.323 is a binary protocol that uses ASN.1 notation/encoding. SIP is a text based protocol that bears many similarities to HTTP. SIP could be implemented in just about any language - certainly perl, java, tcl, or anything else... whereas the much more complex BER encoding (ASN.1) buys you into much more complex code that typically falls into the C/C++ area.

H.323 is the "dominant" protocol now because the intial players jumped in behind it and made it the de facto standard for IP telephony. Now that there is a body of product out there that does H.323, it is perceived as being a requirement for new IP telephony applications that desire to "interoperate." Add to that the fact that the "carriers" adopted it because it looks and smells a lot like SS7, and you have this notion that H.323 is required for serious players wanting to interoperate.

SIP is newer and has a smaller following. Few, if any, SIP devices are out there commercially. I think this is because of the perceived need to support H.323 as the "dominant" protocol. This may, or may not, change over time.

Personally, I think H.323 is far too complicated. There are hideous interoperability issues with it, and troubleshooting them is less than pleasant in the best of circumstances. You need special tools to debug it due to it's binary nature, or have to get very good at reading hex dumps. And, as was mentioned earlier, it has the nasty problem of being firewall unfriendly (compared to SIP).

SIP is smaller and easier to implement. It is text based, and thus requires no special decoding tools to examine messages. This makes it easier to deal with for development, and will lead to easier integration and better (faster, cheaper, and more widespread) interoperability. If nothing else, debugging it is easier. If some new commercial apps would start using SIP we might see more growth in that arena. I think this is starting.

H.323 is here to stay however. It already has a large number of products that support it and those have lifetimes that will persist. Netmeeting alone is enough to carry the ball onward for a long time to come.

I'm glad the OpenH323 project exists - perhaps it will defray some of the pitfalls of H.323. However, as it stands today, OpenH323 is still not functional enough for many companies to consider using to develop IP telephony products... leaving them with the option of paying huge fees to commercial stack vendors for H.323 solutions. Huge fees. Huge enough that the little companies cannot play in this this space since the cost of getting a stack is so dang high. I fervently hope that OpenH323 changes this scenario.

It's too bad that SIP didn't win from the start. Given the fact that SIP is so much less complex than H.323, if the thousands of man-hours of effort put into OpenH323 had been spent doing an OpenSIP, we'd have a solid, reliable SIP stack available today and probably running in hundreds of applications around the world... and H.323 might not be quite so dominant. In fact, if SIP had that kind of penetration in the market then Netmeeting might even support SIP. Then we would not be having this discussion at all.

Sigh. Well, kudos to all who have worked so hard to make OpenH323 a reality - we're on the path now, and cannot abandon it. I wish I had time to help more with it.

Greg

/********************************************************************
Greg Herlein Quicknet Technologies, Inc.
Member of Technical Staff 415-864-5225 x541
gherlein@quicknet.net http://www.quicknet.net
*********************************************************************/

-----------------------------------------------------
OpenH323 mailing list - openh323@openh323.org
OpenH323 home page - http://www.openh323.org
Maintained by Equivalence Pty Ltd - http://www.equival.com.au
Using majordomo 1.94.1 - for information on commands,
send email to "majordomo@equival.net" with the text
"info openh323" in the body of the message
-----------------------------------------------------

Links

Columbia's SIP site

Comparison of SIP vs. H.323 by Jonathan Rosenberg (PDF)


back

Valid HTML 4.01 Transitional