Wednesday, July 09, 2003

Telirati Newsletter #3

In my third newsletter I had I grown to take a skeptical position as the core of the piece, providing the reader with a gauge for measuring the viability of a technology that has long held promise, and still does. Again, however, we see that hype of a 1997 vintage is still being poured today from new bottles.

As with my previous retrospective, it's still hard to say what to buy and for what reason, in CPE or in infrastructure. The vitality of VoIP all comes from alternative applications like voice chat and the implementations of push-to-talk in non-iDEN mobile networks (which is an intriguing example of how an application like voice chat that was deemed too low-quality to warrant notice by telecom people comes back around to influence a significant telecom technology).


Newsletter #3: IP telephony is here when…

IP telephony is coming, no question. But the real question is "When is it close enough to my products that I better take it into account in my next product generation?" Not everyone is making IP telephony products - products that enable IP telephony directly. But every telephony product will be affected by IP telephony. So how do you get a weather report? What do you watch to tell how close it is to your market and your product? Here are some things to look for on the horizon:

IP telephony is here when…

A major wireless service provider builds an infrastructure based on IP telephony.

Wireless service is at the wave front of providing telephony where is has been non-existent or sub-par. The newest infrastructure, in the most challenging environments, is wireless. By a "challenging environment," I don't just mean a jungle or a desert, I mean countries where the per-capita GDP is less than your cellular bill for a month. Cost and price are as big a challenge as snakes and alligators. If aggressive use of new technology cuts costs, wireless providers will try it. A key indicator of the readiness of IP telephony will be the use of IP telephony to reduce the cost of a wireless telephone infrastructure. Alternatively, you might see IP telephony in a wireless infrastructure geared toward providing Internet access as a key selling point.

A significant long distance carrier runs a major part of their network packet-switched.

IP telephony is the new frontier of the brigands of telephony - callback operators. They have decided to hoist the Jolly Roger and sail out in pursuit of the domestic telcos' galleons, having brought home enough booty from overseas PTTs to make them bold (and hungry for the next big kill). Their entry into this market will test the technology of IP telephony. It will also test a number of other things - all of them important markers that show how real IP telephony is: Can they runs these networks with acceptable reliability? What will billing look like in the Internet telephony domain? Will corporate telecomm buyers use these services as readily as they use callback overseas? Can these buccaneers with their risky technology learn how to market themselves as respectable?

You can buy a $200 telephone that runs IP telephony.

The IP telephone is really a set of solutions to a very infrequently mentioned problem in IP telephony: the terminal. Telephones are good voice terminals. Computers are not. Telephones are an accepted and well-tuned part of how people in a business setting communicate. Computer multimedia devices are not. My mother can use a telephone. I won't hold my breath for her to use any current IP telephony technology. What the world needs is both a stand-alone device that connects to an Ethernet jack and provides what looks like normal telephony to anyone who lifts the receiver, and a multimedia i/o device for computers that is as nice as a good speakerphone handset from an established CPE maker. Some makers of network telephony solutions enable users to plug a phone into their expansion board, which is installed in a PC. Windows can manage multiple audio devices reasonably well, and this is about as well as can be done at this point, but it is a far cry from being as good as those really nice speakerphones Nortel makes. You can't make a 2500 set hooked up to your computer auto-answer, you can't use the programmable buttons or the "hold" button as you would normally - it just isn't like using a phone. And then there is the problem that opening that many PCs and installing that many boards is a non-starter. Ordinary modems, though almost all have both a phone and a line jack, cannot be an Internet telephony i/o device - that jack exists only to make the phone usable on the same line as the modem. This is not a minor quibble: I can get PC multimedia devices and telephones at every office supply computer store, and consumer electronics store in the country. I cannot get a device that serves my PC multimedia and telephone needs in one package anywhere. The solution to this problem is interdisciplinary and difficult. But lacking a solution to this problem means IP telephony is not ready for prime time.

A call center, for both Internet and telephone-based callers is built in the IP telephony domain.

Here is a nicely constrained problem: You are building a call center. Everyone has a PC on their desk. Everyone wears a headset all day. Since it is a new call center, you are quite free to specify audio cards for the PCs that are well suited to Internet telephony, as well as every other attribute of those PCs. You have control over the network. You know exactly what will be moving on that network - you can calculate worst-case traffic down to the bit. You choose the NICs, the hubs, the switches, the routers. Why then should you not do away with the cost of a call center switch, switch-to-host integration, etc. and go straight for IP telephony for handling all calls, whether they come from the Internet or the PSTN? How convenient for your server software to talk directly to the IP telephony software, inside the same server, or at least inside the same OS on the same LAN, with all the OS's inter-process communications capabilities easing the way. One OS, one wire to each desktop, one network, one management system, one software technology for implementing every component - all very attractive, and all likely to save you a bundle. Of course, if you were really doing this, there would be hell to pay if it did not work. So it makes a nice barometer: when you see someone stick his neck out and build a call center this way, IP telephony is on your doorstep.

Copyright 1997 Zigurd Mednieks. May be reproduced and redistributed with attribution and this notice intact.