WARNING: This is a long, boring and esoteric post for the most part. All except for Dr. Levy, it is at least hoped. Therefore you travel here at your own risk!
==========
Bernard,
Re: Your post #1156 that dealt with Professor Noll's voice versus data argument
I appreciate and respect Professor Noll's spirit and the nature of the problem which he attacked, but to be honest with you I have not been able to determine if it was an altogether serious effort, or one which had as its central focus some kind of academic amusement.
>> Such claims are not new. As long ago as 1961, Frederick R. Kappel, then chairman of the board of AT&T, predicted that data traffic between computers would be greater than the communication traffic between people within 15 years.<<
Yes, I remember this period in time (the latter part of the Sixties) and the issue of voice versus data. It seemed to be an obsession with some, and it was the medium of some blatantly territorial intrigue, anthropologically speaking. Believe me on this one.
>> In the article titled "Future Data Traffic Volume," published in the September/October 1970 issue of Computer magazine, Roger Hough, then at SRI, showed the absurdity of this prediction by carefully extrapolating data and voice traffic over the ensuing decades. <<
In 1970 Roger Hough, if it's the same man I knew back then, was president of AT&T Long Lines, and sat in an office about two stories above mine at 32 Avenue of the Americas, I believe, not at SRI. Maybe it's another Roger Hough who also likes to talk about bandwidth utilization things, don't know. <scratch scratch>
Hough, or anyone else for that matter, could not have even imagined much less accurately extrapolated the extent of data traffic which would have been around in the Eighties, or voice traffic for that matter, since fiber had not been tapped into yet and existing forms of media were way too constrained to grow accordingly. Much less that of voice and data during the current period of the Internet Nineties.
He notes:
>>Voice communication between people measured in bits far exceeded data traffic between computers. Yet claims to the contrary are still being made today, bolstered by the success and hype of the Internet.<<
[[Here we go with a digression: I spent the better part of two years of "on line time" between '95 and '96 attempting to reach what amounted to an assessment of an almost surreal quantity. The goal was to determine the total amount of bandwidth that supported all forms the traffic in the US, all of the states that bandwidth existed in, and ultimately the total national bandwidth of the US. This was at a time when WDM was first being introduced onto the WAN scene.
The latter was the infamous Total National Bandwidth thread I've mentioned here in the past. TNB, for short. It was an excruciating exercise at times that consumed several dozen participants' time on the Compuserve Telecomm Forum, and an exercise that ultimately proved to be a road to both infinity, and nowhere, at the same time. But what an education it was! End Digression.]]
The Internet is just one of many different clouds that have thus far grown to gargantuan proportions, relative to yesterday's networks. You may have read one of my previous posts here in SI that talked about a pending consortium effort to bring all of the nation's banks into a single image-based clearing and settlement system. Such a network itself is in its formative stages, undergoing proof of concept, and would rival the capacity of the Internet, if in fact it does not become a tenant of the Internet, first.
But by that time the internet will have lost the characteristics that have made it what it was up until now, which is primarily an open and fundamentally shared universal resource. When you start to place prioritization and policy-driven rule-based restrictions in a quasi-permanent, or virtual private network, form of infrastructure, you are no longer on the Internet of today or yesterday. You are on something new. And this is the end of my second digression. <grin>
I was in international operations at AT&T at the time that the above referenced application utilization "contest" (data versus voice) was taking place (circa 1966 thru 1971) and I recall vividly the tracking of what forms of traffic were ahead of what-all at the time, in the international sector. I wasn't concerned with domestic traffic then, but I knew at the time that it was a matter of interest to the domestic private line folks and the Class 4 switching staff, as well. Which was going to exceed which? and by how much? and by what point in time?
At the time, voice was still in the lead, obviously, but data was indeed catching up. In fact, I seem to recall at least three times in the past thirty years that data almost caught up to voice. I think now it's surpassed, and will stay that way. Strange, wouldn't you say? And what will voice be counted as when it rides here in the form of IP traffic. Will it still be voice? or Data?
Consider, though, that the highest line rates for data achieved on analog voice circuits then (during 1970) was still in the bleeding area of 600 baud to 1200 baud, with 1800 and 2400 just coming out of the hatch "for government applications," and those circuits had to be "conditioned" for all sorts of exotic parameters.
Bits were not used as a yardstick then, since in those days there were none for voice. All long haul traffic took the form of FDM carrier systems known as L Carrier, R Carrier, etc. PCM and T-1 carrier digital systems were in their infancy still, relegated at the time to short-haul trunking "spans" between adjacent central offices.
The long haul stuff was still mostly coaxial systems then, measured in Hertz, not Bps. I.e., traffic wasn't measured in bits, rather in the number of ~ 3.7 KHz channels being used. Channels were rolled up into groups, which were in turn rolled up into master groups, which were grouped further into super groups, jumbo groups, etc. At the top of the heap there were some 3,600 voice channels on a single coaxial "tube" and that was considered really booking it at the time.
I couldn't help thinking at the time that this form of speculation of voice versus data was turning out to be a favorite pastime and topic of discussion for carrier folks. Sides were drawn during coffee breaks, and it was like the old Dodgers versus Yankees thing here in NY, where voice people defended the lead that voice had had over the years, and the data (and primarily teletype!!!) people were threatening to overcome and take the lead very shortly. You see? Some things don't change in this regard. Tomorrow it will be an argument whether voip should be counted as voice or data. And so it goes...
But things were a lot different then, since, at that time, a ledger could have been used to post all of the channels of record, and a tally could have been made (and in fact, was made), however tedious that was to accomplish, and it could have plainly shown how many channels were used for voice and how many for data.
But that was then, when there was one Bell System that carried better than 80% of the nation's traffic, and every central office had a running tally of every type of circuit riding over what specific kind of channel in their control. And the independents were similarly managed, and they could have contributed the same kind of information, and in fact did. So, keeping score in those days was very simple, in comparative terms.
The author notes:
>> A few years ago, a public document issued by Pacific Telesis claimed that data traffic exceeded voice traffic. I called Pacific and asked for the details that supported this claim, but no details were ever given to me.<<
I wouldn't have answered him either a few years ago, except to inform him of that fact. The alternative would have been to enter into an accounting exercise that could have taken years to prove out.
>> Also a few years ago, I calculated estimates of the traffic in bits from a wide variety of services. Nothing came even close to the traffic carried over the telephone network. The published paper presenting the results of my estimates ("Voice vs. Data: An Estimate of Future Broadband Traffic," IEEE Communications Magazine, Vol. 29, No. 6, June 1991) was criticized, yet no evidence was presented to support claims that Internet traffic exceeded voice traffic.<<
I can guarantee you one thing: The author didn't have a grip on what was real or fake, or even out there at the time, where it was, or who was using it. Today the problem would be ten-thousand fold more complex.
The banking industry's bulk data transfers alone, and maybe helped by credit card verification lines and ATM Machine lines, and lets throw in a few on-line trading systems while we're at it, would have been a bear to calculate, if in fact one could even get their arms around these services. Much less quantify the number of cumulative "bits" travelling over them. Oy!
I think that the next paragraph is something that's really special:
>> I wondered whether my estimates were outdated and wrong. So in an attempt to estimate orders of magnitude, I asked my students to estimate the per-day frequency of their telephone calls, e-mails, downloading of files, and Web surfing. I then converted this information into bits. The results confirmed my views that Internet traffic continues to be vastly exaggerated.<<
After reading the foregoing, I couldn't help but feel a twinge of embarrassment for the Professor, as I would have for a skater who landed on her rear end in front of a wide spectatorship at Madison Square Garden. How naive. To suggest that such a sampling would be relevant in beginning to assess real-world volumetrics of the Internet and PSTN networks? I'm still embarrassed.
One of our clients has an application we're "considering" putting onto an enterprise intranet arrangement right now. It will transport upwards of 110 Gigabytes of daily backups and several hundred meg of ancillary data. And that is because it's not big enough to deserve passage over one of the dedicated hi-capacity links reserved for bulk data transfers. [For anyone interested in this sort of thing, we're going to be recommending Sterling Software's NDM over IP application icw an OSPF backbone ...]
So, what's my point? My point is that this secondary application if placed on line would carry more bits than Poll's entire university's voice and fax traffic. And this is an elective application I'm referring to here, not even something that is imperative.
I'm not going to cover blow for blow the professor's additional findings because I think that it's all the same irrelevant thing from this point out. I am still struggling to discern whether I may have misread his intent, and if in fact I am the naive one, falling for some kind of ploy. If it's not a ploy, then I still find it difficult to follow how someone of Noll's stature and background (I visited his web site and was impressed with some of his other areas of interest) can begin to evaluate the magnitude and breakdown of a global telecommunications infrastructure on the basis of what a handful of students do over their phones, keyboards, and fax machines during the course of a day.
I found the entire article somewhat incredible, but entertaining, and evoking of memories from another era, nonetheless. Thanks for sharing it here.
Best Regards, Frank Coluccio |
