| QUESTION & ANSWER |
Cerf: Enterprise VoIP requires cautious approach |
 |
By Amy Storer, News Writer
05 Apr 2005 | SearchEnterpriseVoice |
|
|
Vinton G. Cerf, the co-creator of TCP/IP and senior vice president of technology strategy at MCI Inc., prefers to avoid what he calls the "current VoIP hype" because, in his opinion, it's just one among many services available via the Internet. Still, he like many others recognizes that VoIP is changing the Internet forever.
The industry luminary and ICANN chairman spoke with SearchEnterpriseVoice.com about VoIP's emergence, long-term effects on the Internet and the rapid pace of implementation within the enterprise. How is VoIP changing the Internet?
Vint Cerf: Those of us who were involved in the early days of Internet development were experimenting with VoIP as far back as 1975. In fact, IP was separated from TCP somewhere around 1978 to specifically support low-delay applications, including voice. So, this is nothing new -- conceptually speaking -- there just wasn't enough network capacity at that time. Fast-forward thirty years later, the [global] network now has a substantial amount of capacity and it's very clear that it's capable of carrying real-time transmissions of voice and video.
Two-way interactions, for which low latency is important, developed over time, but they are more challenging because any variation in delivery is noticeable. People started using virtual private networks, like ATM, and the public Internet to carry voice without asking permission of any of the service providers because they didn't have to. All they needed to do was have software and equipment at the edge of the Net to emit the appropriate packets. This phenomenon has been emerging in part because the users at the edge have been able to wire hardware and software to make sound travel across the Net.
A lot of people think the big news here is that sound can travel across the Internet. But really, the profound thing about all of this is that software at the edge of the Net can reach in and manifest different modalities as opposed to asking for that service from the network switches.
Do you think companies are moving to VoIP too quickly?
Cerf: It's always possible to jump too quickly without thinking about VoIP benefits or obstacles. I wouldn't suggest that a company move to VoIP without thinking about the economics and access capacity. I can't make an absolute judgment call about companies moving too quickly or not, but I am concerned about companies moving to VoIP without an adequate analysis.
However, MCI sees many advantages of doing this, and that's why we're investing in things like our converged packet access strategy, where all the functionalities that were previously separated into multiple networks are being combined on one broadband access network. That saves enterprise expenses and it allows for efficient use of access capacity.
You've said you want to avoid contributing to the hype surrounding VoIP, but do you still think of it as hype?
Cerf: What concerns me is the hyperbole associated with it, particularly in regulatory circles, which focuses on VoIP and acts as if it's the only thing you can do on the Net. I'm not happy with the magnification on VoIP because of the impractical, unnecessary and disruptive attempts to regulate it.
I don't think we should even associate VoIP with traditional telephony and they certainly shouldn't be equated to each other. That is the fundamental misunderstanding about what VoIP really is. VoIP is just one more modality ported by the Internet on a very long list. To focus only on this and imagine that it requires some regulatory oversight is just wrong.
Are government regulatory concerns a reason not to implement VoIP?
Cerf: I hope not. MCI is certainly moving along with it. We believe it should be an unregulated application on the Net, but we're going to implement it regardless because it's more efficient for us.
Should enterprises be concerned about the issues with Emergency 911 and VoIP?
Cerf: Yes, they should be concerned right now. However, it's my opinion that we can do a better job with E911 in the VoIP world than we have done with the 911 in the traditional telephone world. As it turns out, when you connect to the Internet you are typically assigned an IP address using a protocol called DHCP. We can extend that protocol to not only give your IP address, but also your postal address and maybe even your GPS location. That's not something you can do today, and it's much harder to do with a mobile [phone]. In the VoIP world, that information will become a piece of data that's in your instrument, and that instrument will give the emergency services your precise location.
You've said that MCI is preparing to have all of its voice calls travel over its IP backbone by the end of this year. Is that realistic?
Cerf: Yes that is true, and yes it is realistic. However, it's important to note that the backbone we will use will likely not be a public Internet backbone. It will be on one that's more controlled.
Do you see system architectures becoming more or less Internet-oriented over the next decade?
Cerf: I think it's inescapable at this point that architectures will assume that they are operating in a network environment, and I think there's a pretty good chance that the network environment will continue to be the Internet. Because it's architected as a layered protocol structure, and you can get access to the Internet at any of those layers, you're free to build applications that interact with the Internet as you deem to be appropriate.
Is the definition of the Internet changing?
Cerf: The term "Internet" usually implies the public Internet. Although I'm absolutely persuaded that the public Internet will continue to grow for business and personal interactions, there is a very strong trend of incorporating private networking into the programming paradigm. There is no question in my mind that networking is going to be a critical element in future architecture and it looks like the Internet will be the most prevalent technology used.
How well-suited is IP for voice traffic in the long term?
Cerf: It's actually proving to be very well suited. The one place where you might run into a problem is at the very edge of the network, where you might get some competition between a heavy-duty file transfer and a voice interaction. In that particular case, marking high priority packets is helpful. In the core of the network, data rates are going up to much higher speeds and now it's so high that you don't have to do very much to emulate the behavior of a traditional voice system.
|
|
|
|
