VidMid VC Call 25, August 2003

*Attendees*
Jill Gemmill, UAB
Art Vandenberg, GSU
Tyler Johnson, UNC
Nadim El-Khoury, UNC
Tom Barton, U. Chicago
Tarun Abhichandani, CGU
Jeanette Fielden, Internet2
Lisa Hogeboom, Internet2
Ken Klingenstein, Internet2
Steve Olshansky, Internet2

*Discussion*

Tyler has been having conference calls with development and strategy people of H.323 vendors to let them know about H.350. He has been receiving good feedback from them about H.350. The real key to vendor support will be the level of adoption of H.350 within higher education and private industry. Most of vendor community recognizes it as an important technology and if they see it being supported by the customer base, they'll support it.

Tyler also felt there was buy-in of the importance of integrating with the existing enterprise infrastructure. People have built video and VoIP as separate stand-alone systems. Entering information manually is overwhelming them. So they're interested in hooking into the enterprise directory so they can scale up. It's not clear if they grasp all the implications of authentication. While H.350 doesn’t directly address federated models it does set some groundwork. Institutions will verbally commit to H.350 but it will likely take the imminence of an application that needs the functionality of H.350 to drive the actual implementation of it.

If there were an opportunity to support development work for a SIP proxy and client what would be needed? On the H.323 side the ViDe.net gatekeeper will be shipping soon so most of ViDe.net would be potential users. There is work on implementing a SIP solution based on H.350. One option is using the iptel open source code. Porting it to LDAP will involve considerable work.

One challenge of open source development is that some institutions use proprietary vendor VoIP systems that have handsets with extra features, such as message waiting lights, and like having a vendor responsible for any problems with the system. They're not looking at SIP as a dynamic environment. They just want to buy a proxy and not deal with open source. Other institutions want to deploy SIP, and have next generation collaborative technologies all dynamically wrapped up together, and not be as concerned about having a vendor responsible. The emphasis for a project could be that VoIP means standards based, PDA based, PC based, and the proprietary stuff can be shimmed in at the edges.

There will be a new working group called the Internet2 Instant Messaging group that will work on the infrastructure and integration of instant messaging. They will begin with building an enterprise, federated version of Jabber. Once the Jabber work is completed, the group will look at SIP/SIMPLE instant messaging as well.

Microsoft has SIP clients built in and their SIP server is out of beta and in manufacturing. Enterprise pricing for it has been set and it should be available within a few weeks.

Supposing we want to do a SIP proxy would we have to do a client? Jill mentioned that to do any authentication other than digest MD5 a client is needed.

Tarun pointed out in his testing of user agents there were differences in behaviors. He has been testing between Dynamicsoft and Vovida and performance is very varied.
Some phones and proxies go across NATS some don't. Ken explained that the NAT issue came up at the Security at LineSpeed conference. If one client/proxy is behind a NAT and other isn't it works fine. If both are the handshake will fail. The issue needs to be promoted so it's more generally known.

There is also the issue of cost in developing user agents. For the proxy SIP just sets up the call and gets out of the way. The user agents start exchanging media, which requires codecs and the increases costs since there are not many open source ones. Interoperability issues are considerable as well. That might vary if you consider Jon Peterson's draft. There are some issues where the client has to play a role, and some issues where the proxy has to play a role. In the drafts he mentions ways the server might play a role. Jill has contacted Jon and the authentication service draft is making its way through the process towards completion.

The general view was that the proxy was the priority since it ties into the enterprise and creates an imminent application to drive implementation.

Ben Teitelbaum has been working with MIT and Yale on a fall demo of SIP telephony. It would be of great value to have Ben on a call to talk about issues with proprietary features of phones and get a sense of how it is working with the iptel code.
[AI] Ken will talk to Ben Teitelbaum about VoIP work and participating in a VidMid VC call.

Tyler brought up that in talking with a major vendor about supporting H.350 in their proxy, the topic of how it related to SIP.edu came up. The bullet items for getting SIP.edu to work includes: connect your proxy into your directory, but it's not specific on how it's done. If every school's directory is different there is a problem since vendors want to have a standard architecture they can cookie cutter. It would provide standardization and promote H.350 if it were the solution for how you connect your proxy to a directory.
[AI] Ken will talk to Ben about including H.350 as the solution for connecting the proxy to the directory for SIP.edu

Discussions of what will be needed in terms of a budget will be a topic for discussion on the next call.

A BoF session at the Internet2 Fall meeting in October has been requested. For now the topic will be left as Authorization and Authentication with a pointer to the website for a full description. Lisa will let us know when we come up on the printing deadline for information.

Nadim is still finalizing the VidMid VC work plan. He and Tyler will not be on the September 8th call since they will be returning from the ITU meeting in Paris on that date.