Hello, Lucius. Permit me to begin by answering your last observation, first. I'm
assuming that you are referring to the other industrialized nations regular POTS or
PSTN systems, and not their fledgling VoIP networks , when you say:
"Elsewhere in the industrialized world it's often not so good, but users still get
messages through."
While that may be true, VoIP's still-inferior standing in comparison to what we regard
as toll quality here in the states, is actually an 'improvement' in some cases, when
delivered to certain spots of the globe. This is because VoIP delivery "when provisioned
over adequate links" is better than their own infrastructures' capabilities.
This is true for many parts of Russia and China (although China still lacks the necessary
deployment of adequate numbers of E-1 and higher capacity lines), sections of SE Asia,
some of the tropics, and many of the third world countries that are fortunate enough to
even have a regional PSTN/GSTN reach, to begin with. This may be hard to fathom,
but it's documented and true.
Having said that (and, admittedly, the foregoing will run counter to some of what I have
remaining to say), I can start off by saying that a thorough reply to your points would
consume many conversations, not just a simple reply here. A good place to start off
with, in order to get a grip on some of these issues, would be to read the "nethead vs
bellhead" article I recently posted about in the VoIP thread:
"Netheads versus Bellheads -- Research into Emerging Policy Issues in the
Development and Deployment of Internet Protocols"
Authored by Timothy Denton, with Francois Menard and David Isenberg (Isenberg,
being the same Isenberg of "The Stupid Network" fame)
Message 10250599
That article is good for the way in which it catalogs many of the issues surrounding not
only voice, but all Internet protocols, although special attention is given to time-critical
applications such as voice and real time video applications.
---------
"...about QOS of IP. This seems to be a big issue, but my understanding of
the technologies involved would seem to indicate that it will be a long time (if
ever) before packet switched can equal the reliability of circuit switched. The
question is how good does IP have to be to be good enough?"
Plenty good, if for no other reason than to dispel the prejudices that currently exist
against its coming into being in the first place. Beyond this, aural quality is not the only
issue that plagues Internet Telephony and VoIP (there are differences, by the way,
between VoIP and Internet Telephony which I believe are spelled out in the above url...
if they are not come back and I will elaborate further on this point).
Some of these differences are so stark as to cause one to wonder if we are talking
about the same parameters of telephony as we are accustomed to thinking of today. For
example, Internet Telephony purists question the wisdom of billing for telephony on a
per call basis, if, they argue, the service is actually part and parcel of the Internet's
delivery scheme in the first place. Would you charge for every email sent and received?
Or so goes the argument, in its simplest form.
But assuming that you do charge, where are the international agreements in place to
assure that there is a uniform means of settlements? The three or four Internet
settlement firms which were set up recently to do this are following some pretty
in-house developed conventions. They are nowhere near being considered universally
accepted standards that would befit an ITU like guideline, for something like 10,000
service providers to agree on. Purely wild west stuff right now. And the prospect for it
remaining this way is high, given the reluctance that governments have shown in
tampering with the 'net in general (except for India, some FE Asian countries, and a
handful of others).
-------------
Getting back to the QoS issue, though...
"Once a "good enough" level is reached, then it can be used where the
price/quality ratio makes sense."
Good enough, where? In the WAN domestically? Internationally? On the LAN in the
office space? Each of these has different parameters associated with them, and they
are not seamlessly joined right now. Yes, it's possible to over-provision a channel with
bandwidth, and create private virtual channels or RSVP links, and derive a quality that is
actually superior to the PSTN. This point should not be lost, for one of the future
benefits of VoIP (using that term generically now) is the ability to convey high fidelity
and stereophonic/quadraphonic even, sound, in the future. We aren't there yet, by any
means, granted, but it should demonstrate the slope of the curve that is being traveled at
this time.
But even where acceptable links are created on private IP backbones (which is how
most of today's successful VoIP services are being delivered), there is no assurance
that these can be commuted to the public or other private IP backbones without
suffering degradation and interoperability issues. So, even if the "sound" of the voice is
okay, it is not necessarily communicable across disparate network domains, as things
stand today. And there are probably hundreds of thousands of different domains at play
at this time, when you consider all of the enterprise intranets in addition to all of the
public service providers' nets at the same time.
I would have agreed with your following statement about three years ago:
"Especially for voice, many people might prefer a much cheaper service
even if it was less reliable."
Think about that one for a moment. When interstate voice was still 28c per minute,
what you say might have been a no brainer. Today, folks are getting voice below 8 or 9
cpm, and rates being given to enterprises are even lower. Why would one want to save
two cents and suffer the indignities of 30-digit dialing, and lousy sound quality, in order
to save on a couple of pennies. Especially when using the phone to talk with clients and
other important entities where you want to make a good impression?
"Error correcting can compensate in the case of data."
I'm not sure what you mean here. Are you saying that non-voice data packets that
could be repaired through error correction? In that case, I would agree with you,
although I don't see the point entirely.
Or, are you referring to error correction for data that constitutes VoIP? In the latter
sense error correction is not used at all, in most designs. Instead, dropped packets are
filled by predictive and algorithm driven means in a fuzzy kind of way, or left
blank altogether without any noticeable consequence, since the ear cannot discern
subtle packet level losses, anyway.
"The only reason that we have such reliable phone service in the US is that
AT&T was smart enough to figure that customers had no price to compare the
service with so would put up with high prices if service was good."
I don't think that it was entirely as sinister as you suggest. The Bell System dragged its
feet, alright, but no slower than the remainder of society. In fact, I'd have to say that
they were ahead of many of society's other sectors by many measures. And during
their feet-dragging, they had a long time (okay, it was a leisurely and relaxed pace...
characteristic of a different era, and without the urgency imposed by Internet Time) to
perfect a platform that we have all come to enjoy today, opportunistically, or otherwise.
Minus this last editorial comment, I hope that the rest of this post has answered at least some of your questions, and helped, somewhat.
Regards, Frank Coluccio |
