Joe Mele On H.323 Version 2 No-Cost
Access I
recently had the opportunity to speak with Joe Mele of elemedia regarding the company's
decision to offer the H.323 Version 2 Protocol Stack to the industry at no cost. Another
important part of that announcement is the addition of the IETF RTP Version 2 software,
developed by Bell Labs, into that free distribution plan. The following is a result of
that conversation. I'd also like to thank Chris Pfaff of Lucent Technologies' New Ventures
Group for his help in setting up the interview with INTERNET TELEPHONY.
--Greg Galitzine, Executive Editor
IT: What is the context of your strategy with this free distribution
of your H.323 Version 2 protocol stack? How did you get the IETF RTP code into this
distribution?
JM: The significance of the announcement comes back to Bell Labs. From
the very beginning, Bell Labs has had many people who work in the standards bodies. Much
of the technology that elemedia employs within its products originates from Bell Labs. We
have feeders from Bell Labs into what we're doing. The IETF RTP code is part of that Bell
Labs linkage. Henning Schulzrinne, the principal author of RTP, has been affiliated with
Bell Labs for a number of years. And, the IETF is one of the standards bodies that
elemedia works with regularly. Jonathan Rosenberg, with Bell Labs, is the chairman of a
new working group in the IETF called iptel. One of the friendly agreements that Jonathan
made with his IETF colleagues was that we would distribute the next version of RTP to the
community for free.
This version of RTP is a research version, which includes the standard as well as some
pre-standard features. When you make this technology available at an early stage, it is
easier to evolve the technology and shorten the development cycles from concept to
commercialization. By getting the early research into the field, the market has a solid
clue as to what the research community is doing. We, of course, were interested in not
only getting the IETF RTP out to the market but also our H.323 protocol stack, so that we
could help speed the kinds of next-generation multimedia developments. Feedback from the
marketplace has identified that the cost of acquiring these technologies has made it very
difficult for innovators to access the technology during the exploratory and early concept
phase of research and development. Both the very large and start-up companies are
experiencing this situation. We at elemedia are trying to address their needs.
IT: What are the means of feedback when you do this?
JM: When you take some research code, it goes into the field and you
get instant feedback from researchers and developers. They give this feedback directly to
Bell Labs researchers. This feedback provides an invaluable identification of which
features are important and/or require modification for commercial use.
Accordingly, this pushes the software commercialization process further upstream to
effectively shorten the overall development cycle and improve our ability to get the right
product at the right time to market. That's the first wave of feedback. The second is when
we you provide access to the elemedia H.323 protocol stack commercial software. We are
making our H.323 Version 2 protocol stack available to the market at no cost not just to
give away code for industry recognition and spur innovation, but also because it helps
improve our product by having feedback from the diverse industry. We believe the needs and
usage of this technology today does not necessary represent the needs of tomorrow.
Availing this technology to the innovators is one method of obtaining a view of tomorrow's
needs - directly from the developers who are creating tomorrow's products and offerings.
IT: So, the first channel is to researchers -- at no cost -- and then to
customers?
JM: This is correct for the IETF RTP Limited Research Edition
software. With regard to the elemedia H.323 protocol stack, the primary target is the
customer or developer, providing access to technology to create tomorrow's products and
services.
IT: What is the feedback loop here, in terms of time?
JM: The entire process is automated and staffed to address and support the demand. The
response has been overwhelming, which validates our assumptions of the market. Our Web
site has been redesigned to accept feedback electronically from "It's easy to
install" to people asking other questions. There's also a second and more personal
method by which we obtain feedback. Pursuant to our agreement, there is a clause that
says, "We'd like to have feedback discussions with you" and we provide a generic
outline for discussion/feedback and periodically contact the customers. This also opens
the door for the customer to more frequently contact us down the road once we have had one
of these feedback discussions. This feedback provides us with a "bird's eye
view" of where the market is going and what will the product needs of the future.
IT: What's the function of RTP?
JM: In addition to media packetization, RTP provides time stamps and
sequence numbers. These are needed to reconstruct the media stream in cases where the
media packets arrive out-of-sequence. Out-of-sequence arrival is possible because all
media is sent using UDP/IP, which is unreliable. RTCP provides for status reports, which
tell the sender of the packets how well the destination is receiving the packets. This
allows the sender the option to renegotiate to a lower media data rate if reception
quality is poor. Many people unbundle the protocol stack and use RTP separately to
optimize its performance where it is needed, such as running it on the DSP assist board
instead of the host processor.
IT: So, you move packetization onto the board side?
JM: Yes, a few systems move this function to the DSP assist board side
of their architecture to minimize delay, which could be caused by host system overhead.
The Limited Research Edition of RTP has all the features of the standard and proposed
draft going in - new algorithms for congestion, traffic, etc. - even while the Internet
draft is going forward.
IT: If you see things that don't work, what do you do?
JM: The developers of the IETF RTP receive market input directly. If
there are issues with the given implementation, they will be assessed and addressed in
future releases. If there are issues with the standard or proposed standards features,
this feedback will be brought back to the working group.
We're finding out that this technology is not the cheapest stuff around and it is not
effective for everyone to create their own implementation. We have previously availed easy
access to the H.323 technology to various development groups. We find out that those
groups truly appreciate this access to the technology and come up with innovative
commercial products sooner. What people do when you give them a technology - and we have
developed carrier-class technology here - is all within the space of innovation and
development for the broader telecom/Internet market.
As I previously noted, the development budgets for both large and small companies can
be really constrained for developers in the early concept phases of development limiting
access to key technologies. We are working with the industry to address this situation. If
innovation can be spurred, commercial prototypes can be worked out. Then funding can be
secured in the organizations. We didn't want the cost prohibiting the many from getting
access to what we hope will grow the overall H.323 software market. We are enabling the
next wave of innovation, which will address the broad adoption of Internet multimedia
communications.
IT: So, if the market grows more quickly, you have a better near-term
opportunity?
JM: Right. As you build this, people build new innovations. You help
grow the market. There is a degree of loyalty that stays with that. There's a commercial
side to everything. I must admit to you, I'm amazed at the application software that has
been created with this technology. These ideas are explosive - you can see exponential
growth percolating from this kind of interchange. You never know what creativity can
emerge from the next download of this software.
|