Response to "SIP and H.323"

From: []On Behalf Of Paul E. Jones
Sent: Sunday, May 13, 2001 1:17 AM
Subject: [h323forum] H.323 and SIP comparison

SIP Center Editor,

Reading this page:

I have to take exception to many of the claims: they're just plain wrong.

SIP does take a simpler view of things, but that's actually a short-coming, because communication systems are complex. If they were simple, we'd create a simple protocol. H.323 is more complex, because it meets the requirements (and I emphasize requirements) in order to build carrier class solutions.

IETF vs ITU?? So?

SIP is carrier class and H.323 is not? Hardly: H.323 is being used for millions of minutes per month in carrier networks.. I don't have the exact stats, but they're quite large and growing daily.

Also, H.323 can use URLs, DNS, etc... it can do everything SIP can do and more. Also, H.323 lists standard codecs that may be used, but it does not limit one to those codecs. Also, for transport, it uses the same RTP/RTCP as SIP.

What is this "leaves issues of reliability to underlying network"? And H.323 does not? It uses TCP or UDP for signaling, just like SIP. What's your point here?

Text vs Binary: "Binary format doesn't sit well with the internet - this adds complexity," Are you kidding? How do you think IP packets are formed? In fact, most IETF protocols are binary. SIP, SMTP and HTTP are the exceptions.

Addressing: H.323's addressing mechanisms scale quite well. I can cite the installed networks as examples. In addition, while most of those networks are using LRQ messages today to resolve addresses, H.323 can fully utilize DNS, TRIP, or anything else-- this argument is completely orthogonal to the signaling protocol. H.323 has an address resolution protocol, whereas SIP does not... and that makes it bad? It's working!

Delay: H.323v2 (and I'll note we are now at v4) has been able to set up a call as fast as SIP. In fact, using UDP, the number of message exchanges are exactly the same.

Simple versus cumbersome: too hard to deploy? Again, I'll clue you in: H.323 has been deployed around the world, including carriers for trans-oceanic calls, residential VoIP services, pre-paid calling cards, etc. It is definitely deployable. Not only that, but H.323 can serve the needs of those who want to do LAN-based video/data conferencing-- something that SIP can't even come close to doing.

Services: Ability to "fork" a call: H.323 can most definitely fork a call. All that one needs to do is have the GK "fork" the call signaling. No, there is no description of this within the standard, but it was not felt necessary: The GK (if it chooses) can have complete and total control over the call signaling and may ring multiple phones simultaneously if it so chooses.

"User Profiling" -- and how does this relate to a call signaling protocol?

"Unified Messaging" -- ditto (there are H.323-based unified messaging systems)

"Presence Mgmt" -- ditto (though we are adding extensions for presence now)

Mix media -- ditto (and many H.323 systems provide IVR)

URL: H.323 has been able to use any URL since V2. We defined an H.323-specific URL in V4, but there's no reason why one could not use the "tel" URL or any other today.

H.323 has trouble connecting to the PSTN? Well, that's about the dumbest thing I have ever heard. Call Beijing, China right now and there's a good chance that your call will be carried via H.323 and terminated to the PSTN. Or try using DialPad or other such services-- again, you'll be using H.323 to terminate that call to the PSTN.

Services: One can define any service that one wants. There are standard "services", but there are no limits placed on implementers. And what is this comment about "voice-only ceiling"? H.323 is fully multimedia capable, including voice, video, and data conferencing.

I'd say that they're both "industry endorsed". The fact that H.323 equipment still out-sells SIP (by leaps and bounds) would suggest something about endorsement.

Vendor support: there are quite a few H.323 supporters out there. Most don't make noise, because they're too busy building and selling products, as opposed to promoting vaporware that's not going anywhere very fast.

Now.. with all that said, don't think I do not like SIP. Quite the opposite: it's a good protocol. But your comparison is completely ridiculous.

Best Regards,
Paul E. Jones