H.323 and SIP

[seifran]

I have been deploying Tandberg and Sony codecs internally on IP, but there is a debate that goes on in my company. The feeling among the engineers here is that SIP will take over from H.323 in video (as it seems to have already in voice), so they are reluctant to take the endpoints I am deploying seriously as none of them support SIP.

The traditional video appliance manufacturers (Polycom, Tandberg, Sony, etc.) have been quiet about this, although the sales and technical types I talk to say that SIP is not enough of a standard yet to be taken seriously. Tandberg is adding the capabilty to work with the Cisco SCCP protocol on the Call Manager, but this only gives a voice over IP compatibility to my knowledge.

Anyone heard or seen any information about this?

[Entropy3XD]

I know that Radvision is developing a SIP solution for integration with H.323 and H.320. I have seen it action and was quite impressed. It is a work in progress, but setup did not seem to difficult.

[Jamz67]

SIP videoconferencing gear isn't even compatible with other SIP videoconferencing gear in my limited SIP experience. So I guess I'm agreeing with the thought that it isn't much of a standard yet. Just when everyone said H.323 was outdated and going away, we're seeing increased use of H.323 videoconferencing products.

My 2 Cents,
James

[Glen Sykes]

SIP as a videoconferencing protocol has a great deal of compelling reasons why it will take over from H.323. One reason I hear a lot is that it is much simpler to implement than H.323, another reason is that SIP is also much more extensible.

On the other hand I hear that H.323 is a 'tired' protocol, and difficult to understand.

But lets take a step back for a moment and look at the situation.

H.323 has been around for a long time, the first time I encountered it was in 1997. As a protocol developed around the telephony system, it made absolute sense to use this system, given that ISDN based videoconferencing would continue to be the dominant method for years to come. To make interconnectivity between the 2 networks simple, H.323 had to grow in the way that it did.

SIP on the other hand has arisen from a more web oriented viewpoint, and issues such as NAT traversal are not as much of a problem. However the problem with SIP is that it is very young, and with it being such an extensible protocol, there is no guarantee that one vendor will develop their SIP offering in co-ordination with the others. This generates the need for a standards body such as the ITU. Inevitably, what will happen if SIP for VC is to be adopted as a standard in the same way as H.323, it will have to be standards based and governed just as strictly. Suddenly a lot of the initial appeal of SIP being simple and extensible goes away.

In summary, I do think that SIP will be adopted, but those who defer their decisions for implementation, especially where there is a business need for videocommunications, should seriously consider H.323 as the means to fulfil their requirement, as I don't think that it be overtaken as the dominant IP comms protocol for a good 2 to 3 years. The other things to consider are that there will be SIP > H.323 gateways, and we have yet to see how IPv6 comes to play in real time comms too.

Glen

[trapehzoid]

SIP as a videoconferencing protocol has a great deal of compelling reasons why it will take over from H.323. One reason I hear a lot is that it is much simpler to implement than H.323, another reason is that SIP is also much more extensible.

Simplier.. mainly in that its less complete and therefore less complex. Its named 'simple' for a reason :) The openess of it is part of its problem.. its too loosely defined.

On the other hand I hear that H.323 is a 'tired' protocol, and difficult to understand.

Its 'mature' now.. through about a decade of revisions. Its not difficult to understand at all.. it just has higher minimum expectations then what all developers may want to implement.


SIP on the other hand has arisen from a more web oriented viewpoint, and issues such as NAT traversal are not as much of a problem.

That is incorrect as well.

I think the adoption of SIP will not come because of functionality or because 'SIP is better' it will come around because of (hopefully) converging technologies between the carrier level providers and business level communications where things like universal access, dial plans, etc all come together. SIP is thought to be 'cooler' and is the buzzword. The reality is its better for call processing servers because its 'lighter' then H323. For instance, the number of ports required per connection is about 1/4th for SIP compared to H323. Thats a big deal for a call processing server. Its those types of reasons, and lesser things like human readable vs ASN.1 why VoIP people like SIP over H323.

The move to SIP will happen, but not full steam until the functionality people are used to in H323 is available, and general interoperability is achieved. This will take years.

So its my personal opinion that anyone making purchasing decisions right now based on future SIP plans is really hurting themselves, because

1) real SIP deployment is far off
2) full featured SIP deployment is even further off
3) Why buy a product today that may do SIP.. but when you are ready for it, it can't do what you want in SIP because its 3yrs behind the functionality you are after.

Like Glen said, there will be gateway and protocal converters available (just like there is for IPv4 to IPv6, etc).

So buy the technology that will work for you NOW, and for the forseeable future. 2-5 years from now when you have to address SIP, you can either migrate your older platforms through gateways/proxies, and buy current platforms that include the functionalites you need.

Why shoot yourself in the foot today for something that may or may not be an issue in the future, which can be addressed then?

[Glen Sykes]

Hi Trapehoid,

Thanks for your response, although I think that you've uneccesarily criticised parts of my response where I already provide the same criticism in my own. Where I state 'I hear.....' it is exactly that, what I hear. My own viewpoint is summarised a little more clearly below.

For example

Simplier.. mainly in that its less complete and therefore less complex. Its named 'simple' for a reason The openess of it is part of its problem.. its too loosely defined.

This is the crux of my argument. It makes little sense to adopt SIP as a standard within an organisation when there is so much work left to do on it, to even bring SIP to the same level of functionality as H.323.

You are absolutely correct when you say that H.323 has a much higher minimum expectation that what what developers may want to implement, but the reason why H.323 has become so sophisticated is because it does not address what developers or system admins want to implement, it's because it delivers features that the end user wants. H.239 is a great example of this.

To Seifran, if your colleagues are not taking H.323 seriously, then I would suggest that you point them at all the major vendors of VC and look at how SIP support is being developed. It's only this previous week where Tandberg have made any official announcement about their endpoints being 'SIP ready'. To delay the serious implementation of VC in an organisation that requires it, because there is a different protocol somewhere on the horizon would be a bad decision.

On the subject of SCCP, The Tandberg 1000 and 550 both are now available with SCCP support, although these products are currently only available (in the UK) to resellers with the Cisco IP telephony specialisation. The SCCP Tandberg endpoints do support video and audio, and the menu structure is much more telephony oriented (such as call hold and transfer on the quick keys). Features such as duovideo are not available in SCCP. I fully expect to see sometime in the medium term a video telephone for SCCP either wholly or jointly developed by Tandberg.