All,
This is a transplant from the Cisco Thread, out of deference to their normal agenda.
---------
To: mindmeld
From: Frank A. Coluccio
Sunday, Apr 26 1998 2:35PM EST
mindmeld,
>>Do you think voice over IP will have the requisite quality and be reliable enough to supplant the current telco-grade switches?<<
That's a fair question. Answering it requires that we think in evolutionary terms, taking things in stages. You've touched on two important attributes: Quality and Reliability. I'll attempt to answer these, keeping several other factors in mind, those of design, scalability, timing (to market) and cost.
===
I certainly do think that there will be the requisite voice quality achievable in VoIP deployments to compare with today's PSTN, but not without strict adherence to network design and engineering scruples. Not only do I think that VoIP quality will be comparable to that of today's switched services when deployed properly, but it has the potential to become eminently superior to that which we have become accustomed to.
I say this because high-fidelity, stereo/quad/you-name-it-phonic algorithms can be invoked at will that are not ordinarily achievable over the PSTN. Today's voice circuits deliver, on average, up to a perceived 3 KHz of bandwidth. Through the use of the appropriate, albeit, higher-class-of-service, codecs/vocoders now becoming available (not currently sold as VoIP, yet, rather today they are associated with multimedia apps) along with real-time protocols and RSVP-like invocations (which have been included in the VoIP Standards Draft), CD-like quality can be achieved for phone calls using IP, sometimes mapped to ATM pipes, resulting in the delivery of a perceived 20 KHz+ bandwidth payload. Good Morning!
Sure, this will cost more, naturally. [But think of the then base price for a moment, and then add 25% or even 100% or so to it.] But not significantly more for the h/w and s/w products, which will cross into commodity territory at some point, if they haven't already, and these higher quality services will probably be metered per class of service during demand usage.
This capability is certainly not without precedent in today's world, since T1/fractional T1 and ISDN have, for a long time, allowed this capability through the use of B-channel bonding and MPEG compression schemes. Rather, my point here is that there is the strong potential and likelihood of it's becoming a reality in the VoIP domain, as well, and once implemented, it would be a lot simpler to select, use and enjoy than implementing ISDN links for each session. Its convenience and luxury factor, in fact, will foster one of the value added attributes that this paradigm will demand to sustain itself. It would merely be a part of the point and click (or DTMF-selectable) converged environment we keep hearing about. After all, when you call that important client who is evaluating your product or service, would you like to present yourself in the image of the highest quality, or that of a call that sounds like chopped liver?
As relates to reliability, if we are talking about the public Internet (and most private adaptations as well) it is not yet technically or economically feasible to approximate the levels of availability using VoIP as that which is found on the PSTN today. The platforms are still buggy, the operators go home at night, and the end game design is still in a state of flux.
And there will often be opposing and proprietary architectures arming to compete with one another at the marketechture level, rather than cooperating with one another at the operations and support levels. Here I am not only referring to the competing upstarts, but the inevitable resistance that each of them will find from the incumbents (ILECs/IXCs) and *their* competing CLECs, as well. This will manifest once they come down to the serious business of all-encompassing telco service provisioning, as opposed to simply slapping a bunch of disjointed boxes (from a cohesive architectural perspective) in apartment building slop sink areas and common utility closets.
While this seems to be changing through the efforts of various forums and the IMTC, there's an enormous amount of mobilizing to do, and there is already a lot of vested planning and infrastructure to be anguished through by some of the startups and would-be standards-setters, that will have to be written off in some manner or reused in some special applications where they can do no harm. These are the growing pains of players in a fledgling industry trying to play Big Boy among the Giants. A fair characterization, I think.
IMO, the reliability will be achievable, over time, but not without premiums to be paid for SLAs and the like, until the model normalizes itself and reliability once again becomes a differentiating hand out or carrot. Several years at the earliest, seven to ten years (and stretching it as a normal measure of hedging <g>) on the outside, where large, scalable and publicly accessible applications are concerned, on a par with today's PSTN.
In the interim, higher levels of quality and reliability for voice services are already becoming available on enterprise nets and private IP backbones, a lot earlier. But these are proprietary solutions, in the main, and require protocol conversion still before they can interwork with the public sector carrier facilities or other VoIP solutions. Nonetheless, these are highly attractive solutions for saving large sums of money, and will proliferate widely over the next couple of years, despite any progress or obstacles encountered on the public side.
The principles I alluded to earlier are, by and large, foreign to most of the ITSP startups staking their claims right now. I'm not talking about Evslin's ITXC or the like, although they've got their work cut out for themselves too. I'm talking about the M&P ISPs and the garage crowd who have become enamored with the prospect of becoming the neighborhood telco. The principles surrounding voice quality and service reliability represent the caveats, I'm afraid, that many of them will find out the hard way, unless their vendors are able to pull them out of the fire, which could happen, out of default conditions outside of their 'control.'
Unbridled and indiscriminate adaptations of VoIP without proper engineering and tuning can, and will, result in a shambles and a level of cacophony and disorder that could wind up rendering its use untenable. It's largely and most importantly a matter of network design and tuning, tuning, tuning with changing subscribership and service feature mixes. Eventually, self-adaptive tuning and service creation capabilities, hopefully, through intelligent rules-based network design and management tools. Don't hold your breath for this one, though.
Perhaps later, (and as unlikely as this might seem) there may be sufficient 'head-room' on the public 'Net to accommodate VoIP with predictable levels of quality and service guarantees, but not now, and probably not ever without some form of premium to be paid to service providers -- where guarantees are concerned. At some point there will be so many variations in pricing schemes, IMO, with each aimed at different levels of QoS, conceivably, that the term Public Internet will have become an anachronism for all but the most pedestrian of applications. Including the lowest grades of voice services.
HTH, and Regards,
Frank Coluccio
|
