Hi Chaz,
>>... Qualcomm's Viterbi said that voice and data should move on separate tracks to maximize efficiency of both. The Q is trying to persuade cellular operators to do just that.<<
There are two ways of looking at this. One is the proprietary way, and the other is standards based.
In the proprietary model both voice and data can traverse the same backbone bandwidth in accordance with the provider's in-house protocols, such as statistically multiplexing packets containing both voice and data. From the proprietary perspective, anything is possible, and soon. In fact, there are international carriers already using this between their gateways, transparent to users, and everyone is happy.
Even though voice can be made to sound good on such a platform, it's restrictive by its very nature from the start, unless it is used for a private enterprise's own purposes, or in the specialized manner like the one I stated above. I say this because implementing interworking solutions for proprietary protocols on the PSTN with multiple service providers is not a cheap or trivial undertaking.
At every provider's border, when proprietary solutions are used, there must be, by definition, a kludge of gateways and other SP elements, for reconciling accounting, administration, authentication, transmission, routing and management.
Some of the startups currently engaged in VoIP, i.e., those who are providing what amounts to tie-line- like facilities [inter-office trunks] between LEC switching centers, oddly enough, don't know about some of these features yet. But they will learn about them, but quick, when the railroad whistle gets closer, within hearing range. Of course, that whistle will be coming from a top-tier service provider, or carrier, who has taken the time and gone through the expense to do it right.
From the standards-based perspective, I think it's fair to assume that there will continue to be separate infrastructure splits for both voice and data during the near to some intermediate period.
We've seen this already. The most viable of the commercial wireline-based VoIP operators, the so called ITSPs, who are using H.323 standards-based platforms. They are employing dedicated IP backbones between their routing/switching locations for their voice offerings. Those who are also providing Internet [data] access services, are doing so on separate facilities, which may be either logically or physically defined. In any event, they are separate, where contention is the issue in question.
Even here they must still hand off to the PSTN, but they do it now with less grief, utilizing what has amounted to, already, "off the shelf" components and software.
The use of separate paths for voice and data will likely continue, IMO, until prioritization schemes are in place that will ensure that voice and other time-sensitive services are not "bumped" or degraded by the equal weighting now held by e-mail messages and even casual surfing activity.
Over the longer term, when CoS/QoS features are in place on a broad scale, we'll likely then see a blending of the two as a matter of common occurrence.
Of course, there are also operators today who co-mingle voice and data over the Internet, and other common media, while offering their services at rock-bottom pricing levels. Some of them are surprisingly good - "sometimes" - but more often than not they only go to validate why the QoS features are necessary in the first place.
This game is no different from any other: In the early adopter stages you get what you pay for. Later, when it is honed and perfected, it becomes a commodity and you wind up paying less for more.
Regards, Frank Coluccio |
