Private Line Services:
I know that this will represent "ancient history" to some
readers, but here goes.
Beginning probably in the 1960s (before my time there) American
Airlines (AA) had what was called a "customer controlled switching
arrangement", or CCSA with AT&T. Of course, they were the only game
in town in those days. AA eventually had almost 150 locations "on
net".
In retrospect, the CCSA was a predecessor to the "software defined
network" that was introduced by AT&T, Sprint and MCI in the late
1980s.
At any rate, the CCSA allowed a company to lease private lines (full
period, no per-minute charge) between the AT&T switching centers, with
Centrex-like phone lines to the various AA offices. I don't know for
sure, but I expect the switches were the Western Electric 4E
tandems. Since calls were not charged per-minute rates the network was
cheaper to operate than just using direct dial. AA was also one of
the first companies to take advantage of WATS (wide area telephone
service) for calls to about 14 or 15 reservations centers. Foreign
exchange (FX - a type of private line) service was used to serve
cities where AA flew.
In the early 1970s Collins Radio developed the first automatic call
distributors (ACD) for use outside the phone company. Unknown to most
people, AT&T had developed their own ACD for internal use, and
possibly for customer use as well, some time before that. I have a
Bell System "Traffic Facilities Practice" dated May 1965 in my files.
Their "service level" objective was 93% in 20 seconds, or an average
speed of answer in 4-6 seconds.
The Collins (later Rockwell) GVS-750 ACD was equipped for 1,024 ports
and was originally designed for the hospitality industry - airlines,
rental cars, and hotels. AA installed five ACDs in four locations
(Hartford, CT, Cincinnati, OH, Dallas, TX, and Los Angeles, CA. The
previous 14-15 reservations centers were consolidated into these four
locations. A combination of foreign exchange (FX) and WATS lines was
used to route customer calls. FX was used for cities where AA flew,
and banded WATS was provided for other locations. There were numerous
WATS numbers in order to keep them in band 3 or less, due to the
relatively high costs at the time. A full-period (240 hours per
month) band 5 WATS was about $3,500 per month, if my hazy memory is
anyway near correct. In today's environment $0.25 per minute may seem
astronomical, but we are talking 1970s here.
The original design thinking at Collins was that there would be about
a 2:1 ratio between incoming lines and agent positions. From a
traffic engineering standpoint that would be true if there were lots
and lots of small trunk (FX) groups. In actual practice, we used much
larger groups (some up to 100 circuits) so our ratio was more like 1.2
or 1.3 lines (FX plus WATS) to one agent. This left a lot of unused
ports on the ACD.
In about 1976 AA hired some people who came out of the military and
had been trained by AT&T/Bell as part of a "training with industry"
program. They saw the possibility of building a private network of
private lines using the "spare" port capacity in the ACDs for tandem
switching. I came out of the Army, having had the same AT&T/Bell
training a couple of years later than the original guys. When I came
on board in 1978 we were well along with the transition. Of course,
by that time MCI and Sprint (OCCs, or other common carriers) were
offering private line service a lot cheaper than AT&T, so we spread
out our service based on cost. There were other issues, such as
circuit order delays with the OCCs (AT&T did everything they "legally"
could to delay interconnection at the operating company) and some of
the OCC circuit quality was not as good. We had a lot of trouble
early on with rain/fog fade on MCI microwave circuits until we
convinced them to install quad diversity.
At any rate, we got our private network installed, and one of my jobs
was to analyze the traffic each month and make adjustments to circuit
counts. We developed an internal analysis system to base provisioning
on cost/benefit, rather than on the traditional mathematical Poisson
or Erlang B traffic formulae. All of the inter-switch trunks were
four-wire, and we used overflow and diversion techniques to balance
incoming customer calls among the call centers.
We used the FX circuits into the busier cities for "tail end hop off"
for corporate long distance calls. If a call was going to an off-net
location we routed it to the nearest tandem switch and hopped it off
to the PSTN at that point. By about the mid-1980s our network average
cost was running about $0.13 per minute. This was when a direct-dialed
PSTN call averaged about $0.25 per minute. We were also using "time
assignment speech interpolation" (TASI) equipment that would carry 31
voice paths on 16 private lines. (FYI, TASI was invented in the late
1950s to increase the capacity of TAT-1, the first trans Atlantic
telephone cable).
In the late 1980s AT&T introduced the Software Defined Network (SDN)
and AA took advantage of the newest offering from AT&T (Tariff 12, it
was called) and we migrated the entire corporate network to AT&T. The
marketing department was quick to soak up the capacity on the ACDs
that we had been using for tandem switching, so we were still making
maximum use of the ACD capacity.
By the early 1990s I was managing the network and AT&T would
not respond to my requests for certain services -- but they thought they
had a "lock" on our business due to the commitments we had made under
the Tariff 12 agreement. We had enough "slack" in the economics so that
I could move the corporate network -- which I did. We kept the
reservations service with AT&T, which by that time was all based on
nationwide 800 service (the new name for WATS), but I took the corporate
network to MCI's Virtual Private Network. AT&T never invited me to any
"outings" after that, but we saved a pocket full of money by doing it my
way. The network average cost was down to about $0.08 per minute.
I moved on to other work in 1998, but when I was "invited" to leaveAA/
Sabre in 1999 the VPN was still working. I do not know what happened
after Sabre was sold to EDS and the world revolved again.
Private lines may still have a place -- it just depends on your
business. For instance, some alarm companies lease "dry pairs" from
the alarm premises to the monitoring station. They are still useful
where you want to load-balance between call centers, or have to do
call transfers to another location. They are still leased (I think)
by the quarter-mile, based on the airline mileage computed using the
V&H coordinate system.
Regards,
Charles G. Gray
Senior Lecturer, Telecommunications
Oklahoma State University - Tulsa
(918) 594-8433
[TELECOM Digest Editor's Note: Since commercial long distance has
gotten _so_ inexpensive in recent years, I cannot think of a single
instance in which a 'private' network would be more economically
viable these days as an alternative. However, many companies and
other institutions continue with private networks out of, IMO, a
misplaced sense of security, i.e. if the local central office burned
down (its been known to happen) or there were other commotions (again,
also known to happen) "then we would have our private network to fall
back on". But would they really? Time and again, those 'private
networks of any size and/or consequence are routed over the very same
wire pairs and through the same central office equipment as the
'exposed and open to danger' lines of the company's "regular" phone
service. After the May, 1988 fiasco in northern Illinois with the
central office fire in Hinsdale, Illinois, people at first were saying
"Well, thank God I have my cellular phone"; or "We can always use our
private system in the office to reach other cities, etc." I am afraid
that was not the case. Those lines were down and out for the duration
also.
So these days, do private line networks make about as much sense as
'banded WATS lines'? How many ways can one pinch a penny? PAT]
Charles Gray (charles.gray@okstate.edu)