Voice
Communication in Business Volume 1
Essays on telecommunications,
1969-1980
Chapter 7
CO Switching: Is It a CU Rip-off?
As the previous chapter
shows, Centrex was, for a number of years, the telephone company's
preferred solution to the PBX problem, and CCSA was the ideal means
of handling tie-trunks. Both of these approaches offered great
benefits to the telephone company, so the customer was charged a
premium rate for making these benefits possible. Although central
office switching had some real advantages for the customer, perhaps
the most important being the telephone company's reliable power
which permits CO equipment to continue working during massive power
failures such as the two day blackout in New York in July, 1977, it
has problems, too.
After encountering
various presentations by Ray Kraus (a professional engineer who went
into independent consulting upon retiring from Pennsylvania Bell
with 30 years service), all insisting that Centrex on No. 1 ESS was
the only way a business customer should go, I wrote the following in
June for Sept-Oct, 1977, Business Communications Review.
Needless to say, I was not as entranced with Centrex and CCSA as
Ray.
***
For many years, central office switching has
been the telephone industry's solution to the communication problems
of large business customers. "Centrex CO" and "CCSA" have become
part of the language and, unfortunately, they have also become part
of the folklore of the telephone industry. As such, they are often
recommended and accepted on an almost ideological basis that defies
sound business or engineering analysis. As an example, one
well-known consultant insists that Centrex CO from a No. 1 ESS is
the only form of extension switching that a business customer should
consider. Many of us have doubts, however, and even the telephone
industry is starting to rethink parts of its position. Thus a
careful look at Centrex CO and CCSA may not be amiss.
Centrex, a coined word that may be a
contraction of "centralized extensions," identifies PBX service plus
direct inward dialing that bypasses a company's switchboard, and
identified outward dialing for per-extension billing on certain
classes of toll calls. Centrex CU refers to Centrex provided by a
switch on the customer's premises, usually a slightly modified PBX,
while Centrex CO refers to switching in the central office using CO
switching equipment such as No. 5 Crossbar or No. 1 ESS.
CCSA is an acronym for Common Control
Switching Arrangement. CCSA is a form of tie-trunk switching using,
once again, central office switching equipment rather than equipment
on the customer's premises. By taking advantage of CO switch
capabilities, CCSA makes tie trunk switching work very much the way
toll switching operates in the public network.
As we shall see, the major problem with
Centrex, and Centrex CO in particular, is cost. CCSA, on the other
hand, has certain conceptual difficulties in addition to high cost
that must be carefully considered by any communication manager when
contemplating tie trunk switching.
ADVANTAGES OF CO SWITCHING
The most apparent advantage of CO switching
is a saving of floor space at the customer's location. Perhaps more
important, centralized maintenance, taking advantage of automated
equipment in the central office and, sometimes, around-the-clock
maintenance forces, can be of great value. The telephone company can
provide better service for more customers with fewer maintenance
people, and those maintenance people do not have to be scattered
over a large geographical area.
Centrex, whether CO or CU, reduces the number
of switchboard attendants the customer must hire, and helps allocate
and control toll costs by billing some kinds of toll calls by
extension. In some advanced forms of Centrex (Centrex II), certain
"sophisticated" features, typical of those found in almost all
modern electronic PBXs, are alleged to be able to reduce key
telephone equipment.
CCSA permits tie trunk switching for users
who do not have step-by-step (SXS) PBXs. For some obscure reason,
such PBXs as the 770, the 800A, and the No. 101 ESS are not well
adapted to making tie trunk to tie trunk (through) connections. Thus
they cannot be used as hubs in tie trunk networks, and some other
means of tie trunk switching must be provided if a hub is required.
The customer who abandoned SXS too early may be forced into CCSA
when he decides on a tie trunk network.
CCSA also provides a "uniform numbering
plan," which permits a location to be dialed by means of the same
code no matter where the caller is located in the network, some
point-to-point traffic information to help in tie trunk cost
allocation, alternate routing for tie trunks and centralized access
to WATS lines.
Some of these advantages are quite useful,
but two questions must always be answered before any major decision
can be made: can the advantages be obtained in other ways, and what
are the disadvantages that go with the advantages? We will consider
both questions as we proceed.
CENTREX CO
From the Telephone Company Viewpoint
The New York Telephone Company invented
Centrex CO to counter the problems produced by the building boom in
New York City following the Second World War. Moving PBXs,
particularly Steppers, in and out of skyscraper locations,
stock-piling them when not in service, and meeting the surging needs
of new office building construction (32 million rentable square feet
of office space between 1954 and 1960) with per-customer equipment
just didn't appear to be practical. A more reasonable solution was
to put one large CO switch at some central point as in Fig. 1, wire
all the nearby buildings directly to it, and use its class-marking
and routing features to provide centralized PBX-type service.
Since the No. 5 Crossbar CO switch was
provided with Automatic Message Accounting, and since its trunking
capabilities were such that it could interface directly with the
public network, about all that had to be added was transfer
capability and positions for switchboard attendants. Office
buildings could be wired up once and for all, and changes in the'
many user groups served by one machine could be made by changing the
control parameters rather than moving a lot of complex equipment
from one location to another. The obvious desirability of this idea
can hardly be overemphasized.
However, it is a truism in switch design that
matrix size and control complexity increase roughly as the square of
the number of lines served. It is much harder to design a switch for
30,000 lines than it is to design one for 300 lines and install 100
of them. Further, not all cities are as compact as New York; the
amount of copper needed to reach 10,000 telephones may well be less
in New York than anywhere else in the world. Thus Centrex CO may not
always be a good idea, particularly in view of the increasing cost
of copper and the decreasing cost of electronic circuitry.
Another point of more than passing interest
is the growing shortage of telephone numbers. Any number for direct
inward dialing must be a part of the national numbering plan, and
such numbers are in limited supply. It is perfectly possible to
address a 3,000-line PBX with a single number. With Centrex, 3,000
numbers, about a third of those served by an office code, would be
required, even if some extensions did not need direct inward
dialing. Now, after almost 20 years of direct dialing, we find a
number-short telephone industry starting to give serious
consideration to alternatives other than Centrex CO.
Centralized PBX maintenance, in connection
with Centrex CO, can be quite important. No. 5 Crossbar and almost
all electronic CO switches have elaborate automatic troubleshooting
equipment which usually finds impending troubles before the user is
aware of them. However, the office maintenance force may not be
around much longer. It seems that the new electronic offices are so
reliable that maintenance personnel don't encounter enough troubles
to keep their skills active. Rather than put in false troubles to
exercise skills, the telephone company has taken a much more
realistic approach: they are centralizing maintenance forces for a
number of CO switches. One force serving fifteen or twenty
electronic switches via telemetered interactive terminals will
encounter enough troubles to keep up work proficiency. However,
technicians still have to travel to the switch to resolve certain
classes of troubles. They might as well go to a PBX on the
customer's premises as to a CO switch somewhere else.
From the Customer Viewpoint
From the customer's point of view, the
situation is somewhat different. Centrex rates are usually
appreciably higher than those for a SXS PBX. Although floor space
savings have been dramatic for some customers, notably those with
growing needs and no more available space, it often happens that the
actual dollar value of floor space, when the space is available, is
negligible compared with the total cost of the telephone system.
Further, since switching space suffers somewhat from dis-economies
of scale, moving the required floor space to the telephone company's
lease does not necessarily lower the cost. The cost of the floor
space is simply included in the telco bill.
There are some other points in regard to
space. In many buildings, space varies in cost depending on the
desirability of its location. Often, it is possible to put the
switchboards and switching equipment in the "low rent" district
(basement, undesirable floor, closet area, etc.). I evaluated one
Centrex proposal where the cost of floor space actually increased
slightly as a result of replacing a 701 PBX, based on this effect
alone. Because the Centrex consoles had to be located in the high
rent area, the smaller floor space requirement produced no dollar
savings.
When one compares Centrex CO with a modern
electronic PBX, the floor-space argument becomes even harder to
accept. For instance, Executone is marketing a PBX that puts 500
lines and trunks in a 7-foot cabinet about two feet on a side. This
is about the same amount of space as the telco would need to
cross-connect CO cables to station wiring. Although other electronic
PBXs are somewhat larger, they don't need appreciably more space.
This is true of the Dimension PBX as well as the machines of the
interconnect companies.
Reduction in switchboard attendants.
To justify the higher cost of Centrex, direct inward dialing, or
DID, is one of the most important of the Centrex advantages. By
transferring the cost from attendant salaries to the telco equipment
bill, the cost is made more palatable. Further, incoming calls, no
longer filtered through the switchboard, are completed more quickly.
This is an advantage to the telco and the calling parties;
indirectly, it is also an advantage to the Centrex customer himself.
In reducing dependence upon and the cost of
switchboard attendants, Centrex (both CO and CU) provides its
principle benefit to the large customer. A seven or eight position
switchboard can often be reduced to two consoles. However, as the
systems get smaller, the attendant reduction becomes proportionally
less. It is hard to have fewer than one attendant, even if that
person also doubles as a receptionist.
In smaller systems, a single attendant with a
modern PBX can handle a hundred lines fairly easily; two attendants,
sharing the load, can handle a little more than twice the traffic.
Thus, up to perhaps 200 lines, PBX operation may prove more
satisfactory than Centrex. Then, too, there is a minimum number of
switchboard attendants required by a company, based more on company
operations than just number of extensions alone. Robert Townsend,
for instance, in Up the Organization, emphasizes the desirability of
having calls screened by the switchboard rather than individual
executive secretaries. In particular, he suggests, people on the
road, calling in for their messages, can always get to the
switchboard; secretaries, however, may not be easily located when
principals are out of town.
Identified outward dialing. In
contrast with DID, Identified Outward Dialing, or IOD, is often more
useful to smaller companies than to large ones. Large companies can
afford elaborate networks of WATS lines, FX lines and tie trunks to
reduce their toll costs; IOD, of course, does not work on such
circuits. It only works on "dial 9" calls, and in large companies
these calls are usually toll diverted to enforce use of less
expensive facilities.
Small businesses, on the other hand, have to
use the public network, need allocation information and often find
abuse control important to their economic welfare. QZ billing used
to be a good service offering for such companies until it was
withdrawn or priced so high that it went out of reach. In New York
City, the operators who used to obtain station identity or
charge-number for QZ billing are now rendering an identical service
as IOD for Centrex CU, and QZ billing is no more.
There is no particular relationship between
IOD and DID, and many customers need one or the other but not both.
Further, DID can easily be applied to large SXS switches already in
place with minimum investment. Thus, the advantages of DID could
have been made available to most large customers a long time ago.
IOD from a Stepper, of course, is harder—but it is seldom needed,
just as DID is seldom needed by a smaller business with an
electronic PBX and a fast console. The reluctance of the telephone
industry to offer these desperately needed services separately is
hard to understand. But the appearance of the Dimension PBX, plus a
lot of prodding from interconnect suppliers, is starting to make
things change. Since Dimension can do DID about as well as a
Stepper, and since Station Message Detail Recording (SMDR), which
works on WATS, FX and tie trunk calls as well as message unit local
calls, is so readily available, unbundled tariffs are springing up
all over. It is amusing to see the number of Dimensions going in
supported by standalone SMDR recorders.
Reduction in key telephones. In spite
of DID and a reduction in the number of attendants required, Centrex
usually still costs more than a SXS PBX. Thus, additional places to
cut costs have been sought. It was natural that designers,
unfamiliar with business operations, would try to use the advanced
features available from an ESS central office (or any modern
electronic PBX) to reduce the number of key telephone sets at the
customer's location. In a Stepper, key equipment usually accounts
for more than half the equipment cost; its elimination by "features"
would make a feature-loaded Centrex system easier to sell.
Sometimes this approach works. In areas where
the calling rate is relatively low and users answer their own phones
without secretarial screening, modern features can, indeed, provide
the services required. In such a situation, however, key equipment
is seldom needed in the first place. Thus, it's a standoff.
The next stage comes when one secretary has a
call director to answer a number of phones for people who are
frequently away from their desks. These users may very well have
single-line telephone instruments, but the Key Telephone Unit that
supplies hold and lamp control features must be provided on each
line that terminates on the call director. Since the cost of a key
telephone set itself is a small part of a key system's cost,
elimination of the KTU and the large call director can be quite
effective in lowering costs.
Call forwarding on busy and don't answer is a
good feature for this purpose. If the line is busy, or if it does
not answer after three rings, the incoming call can be transferred
to a secretarial phone. Call pick-up can also work here. Pick-up
allows any other user in the same group to answer a ringing phone by
coming off-hook and dialing a service code (often the star button on
a Touch Tone telephone).
But, with either of these features, the
person answering does not know, necessarily, which phone was being
called. To get around this, LSI was invented. Not Large Scale
Integration, but Line Signal Indicator. An LSI telephone set has
status lamps for a number of lines, but no pick-up keys. The ringing
line can now be identified before the star button is pushed for
pick-up. If two lines are ringing at the same time, the person
answering must watch to see which ringing line goes from ringing to
talking status. LSI instruments cost slightly less than regular key
systems.
The heavy use of key equipment comes when
secretaries screen calls for their bosses. The usual procedure is
for the boss and the secretary to share two lines in hunt—if the
first line is busy the next incoming call moves to the second line.
Further, there is a separate intercom between boss and secretary,
with a distinctive signal to differentiate its ring from that of a
regular call. When a call comes in, the secretary hears the bell and
observes a lamp indication on the line. The line is answered, put on
hold, and the intercom used to consult with the boss. The boss
accepts the call by depressing the proper line button on his phone.
If the boss is already busy on a call, that call can be put on hold
by depressing the hold button and then selecting the intercom line
upon hearing it buzz. The procedure is so simple and obvious that
even executives can master it with minimal training.
Note, however, that both the boss and the
secretary have a visual display to guide them in their actions: they
can tell which line requires attention, whether it is idle, ringing,
in use, or on hold. They can select a line (or the intercom line) by
an obvious and simple procedure dear to the hearts of all Americans:
pushing a button. They can park one call, pick up another, park it,
and return to the first call with no ambiguity. Best of all, such
systems are not limited to two lines. A six-button key telephone can
pick up four lines and an intercom; larger systems are available.
The switch features used to try to duplicate
these operations are frequently the
station-dial-transfer/consultation-hold/add-on-conference
triumvirate built into all modern PBX and Centrex systems.
Station-dial transfer is a useful feature by itself; in either a PBX
or Centrex system, it de-loads the switchboard attendant by giving
the user the responsibility for transferring calls. When consoles
replaced cord boards, all calls had to be placed via the switching
matrix instead of bypassing it through cords direct to jacks in the
switchboard multiple. This made the switching matrix larger to
handle calls formerly placed via cords, and made system control more
complex. The attendant had to dial the required extension on an
incoming call and let the machinery take over. On a transfer, the
station user would flash his switchhook as usual to recall the
attendant. The attendant would than release the original connection
and dial in a new one.
If the attendant could dial in the new
instructions, so could the station user. With the outside caller put
automatically on hold, the station user could dial up the new
connection, announce the call or consult with the second inside
party, and then connect all three together for a conference. Hanging
up completed the transfer. In electromechanical systems, these
features were usually built into trunk circuits and only worked on
incoming calls. However, the three features can be seen, in
principle, to be one feature over-all.
To use these features in the boss-secretary
relationship, the secretary's number is listed for the boss. The
secretary answers her phone in the boss's name, flashes the
switchhook, dials the boss's actual extension number, and announces
the call. If the boss wants the call, the secretary flashes the
switchhook and hangs up. (If she doesn't hang up—and there is no way
to tell—a privacy problem exists. Since anybody can transfer a call
to anybody else with such features, privacy is a fairly serious
problem.) If the boss doesn't want the call, he hangs up, the
secretary flashes the switchhook again, and takes a message from the
calling party.
So far, so good. But if the boss is already
busy on his line, he is out of reach. On important calls, the
secretary has to take in a note. Industrious PBX designers have
devised ways in which a call-waiting tone can be applied to the
boss's phone; he can flash and dial a code to put the other line on
hold and respond to the secretary. Then, by flashing and dialing, he
can put the new call on hold while he returns to the original call,
or whatever is desired.
These procedures, while they pose no problems
for design engineers, are bitterly resented and resisted by station
users. Thus it is common to see key equipment, supposedly eliminated
by Centrex features, going back in after a few months.
The communications manager is thus confronted
with the option of providing degraded service to those who actually
need key telephones, or blowing his budget to pay for features and
key at the same time. Either way, he loses.
To summarize, Centrex CO is as good a way as
any, in principle, to switch user extensions. From the standpoint of
the customer, however, its considerable cost is seldom compensated
by attendant reduction and floor space savings alone. Reduction of
key equipment may bring costs into line, but at the considerable
danger of degraded service. If money is no object, choose Centrex.
To the telephone company, Centrex CO appears
to be losing some of its charms. This is doubtless due to advances
in PBX technology, the high cost of running cables, the shortage of
telephone numbers, and, possibly, concession to the needs of the
customer. Unbundling Centrex services, in particular DID and IOD,
greatly reduces the need for switching in the CO where Automatic
Message Accounting equipment familiar to the telephone industry is
available. Better means are now available to customers who want
message detail recording, particularly on WATS, FX, tie trunk and
message unit calls.
Alternatives to Centrex CO
Centrex CO will apparently be replaced by
small, electronic PBX switches on the customer's premises. Like SXS,
these PBXs will be able to handle DID when the central office is
able to send digits toward the customer. However, all extensions
will not have to have extension numbers that match the national
numbering plan. "Selective Centrex" can be arranged so that people
who receive large numbers of calls can be dialed directly, while low
volume extensions can be handled via a common directory number at
the console. Further, since DID extensions can be arranged to hunt
to non-DID extensions, only one DID number is needed to reach the
phones of a two-line executive, his assistant and his secretary.
With a little care, an extension to DID number ratio of about 3 to 1
can be managed, saving money and telephone numbers at the same time.
Station features can be provided by standard
key equipment, standard equipment with conductor-saving wiring
techniques, or new electronic key telephone sets that provide the
best of key and PBX capabilities.. Useful and cost-effective options
are available now, and will be improved in the near future.
The high reliability of electronic switching
coupled with the wiring reduction of PBXs over Centrex indicates
that both customers and telephone companies can, in most cases,
abandon Centrex CO with great benefit to all.
CCSA PROBLEMS
There aren't very many CCSA tie trunk systems
in use; since CCSA, as originally offered, could only be justified
for very large systems. However, the coming of No. 4 ESS and the
offering of CCSA-type services by specialized common carriers
indicates a need for some independent analysis.
From a theoretical point of view, CCSA
suffers the same difficulty found in Centrex CO: a large switching
system, capable of interconnecting many thousands of terminals, is
carefully restricted so that only terminals within limited subgroups
are permitted to have access to each other. This "limited access"
switching in a "full access" machine can hardly be expected to use
the machine's full capabilities and, as a result, increases the cost
of switching.
CCSA has two more immediate problems,
however, and they are even more important in increasing the cost to
the user: access lines and ONALS (off-network access lines.) There
are a few other problems, but these are dominant.
Access Lines
In any form of tie trunk network design, it
usually turns out to be less expensive to put tie trunk switches at
points where most calls originate and terminate.
This minimizes the cost of connecting to
major user concentrations. With a dial tandem network, based on the
701 SXS technology, tandem switches for tie trunks could be added to
the basic PBX with a flexibility and ease that is hard to beat. By
making the tandem switching part of the major PBX itself, local
stations had direct access to tie trunks and vice versa.
Contrast this with CCSA. The CCSA switch is
always somewhere else, and must be reached via "access lines" from
the local PBX (see Fig. 2.) These cost about $38 a month each,
minimum. A 701 PBX switching 500 tie trunks would need no access
lines at all for extension to tie trunk connections (actually,
cross-office wiring would exist, but would not be purchased as a
separate item). Change the tie trunk switching to a separate CCSA
switch, and perhaps 50 access lines would immediately add $1,900 a
month to the network cost. With four or five switching hubs
required, this might add $100,000 a year to the cost of the entire
network. With 701 switching, this cost would simply not exist for
AT&T high-density points.
There is an odd aspect to this access line
business. In Centrex CO, we expect an order of magnitude more wire
between the CO and customer location than with PBX service; this is
the price we pay for eliminating remote switching equipment. But
with CCSA, the situation is reversed. By bringing all the tie trunks
only as far as a CO switch, and reaching the local PBX via an
appreciably smaller number of access lines, the number of pairs to
the PBX is reduced. Using our previous example of 500 tie trunks and
50 access lines, and considering that tie trunks are brought to the
PBX on a 4-wire basis while access lines can be 2-wire, a
pair-reduction of about 20 to 1 is evident when we go to CCSA ... at
an additional cost of $1,900 per month.
Off-network Access Lines
The second major cost problem with CCSA is
off-net access. With a 701, you would simply buy a separate group of
"dial 9" trunks (to separate message unit charges for tie trunk
calls from those originated by local users), put them on the 9-level
of the tie trunk incoming selectors only, perhaps have them toll
diverted, and you'd be done (see Fig. 3). Users at remote points,
wishing to go off net, would dial the code to reach the particular
PBX, dial a 9, and get dial tone from the central office. They would
then dial into the CO as if they were there.
Contrast this with CCSA. Assume you have a
CCSA hub in Los Angeles and a small PBX in San Francisco as shown in
Fig. 4. If you have a lot of business with other companies in the
Bay Area, you'd have to have one group of tie trunks to the PBX, and
another group, ONALS, to serve as FX lines from the Los Angeles hub
to San Francisco. This is because CCSA is incapable of letting you
go off-net through remote PBXs. Where you had one tie trunk group to
San Francisco with a dial tandem system, you must have two groups
with CCSA at much lower efficiency. You'd probably end up with at
least three or four extra circuits.
Why does this problem exist? Simply because a
human can readily identify dial-tone when he hears it, but a CCSA
sender cannot. The user knows when the distant central office has
been accessed, and can dial accordingly. In CCSA, senders are
automatic machines that must contain the entire number to be dialed;
they require elaborate signals from the distant end to know when to
send digits and when not to. These "handshaking" signals are the
conventional on-hook/off-hook supervisory signals arranged in
special ways. But when a distant PBX connects to the central office,
it can't send any such signals. All it can do is pass dial-tone from
the CO. If only Bell Labs could invent a dial-tone detector, CCSA
could be made almost as good as SXS dial-tandem, at least in this
respect.
Uniform Numbering Plan
CCSA has other problems. Consider the highly
touted Uniform Numbering Plan. Such plans usually make you dial a
number of extra digits which would be unnecessary in dial-tandem.
The reason for this is simple. In dial-tandem, you're standing on
the called terminal after the last pull of the dial, but in CCSA, as
in any other senderized system, at the last pull of the dial you
have just finished loading the automatic machinery. How does it know
you are finished? The easiest way is for it always to get the same
number of digits. If the routing doesn't need these digits, or if
the terminal PBX needs only two or three digits, or a terminal key
telephone set needs none at all, extra (dummy) digits have to be
inserted just so the sender knows when it's completely loaded. Then,
of course, when the sender is loaded it must find a trunk, spill
digits forward, etc. This all takes time. The "post dialing delay,"
after you have dialed the last digit and before the connection to
the called line is complete, is appreciably longer than in the slow,
old fashioned Stepper. The latter is already finished before CCSA
has even started.
Alternate Routing
Alternate routing is another difficulty.
Alternate routing can save you a good deal of money, even in small
networks. If you can arrange for one or two (or more) direct tie
trunks from one location to another, and keep these direct circuits
fully occupied by letting all overflow traffic be routed through a
tandem hub (see Fig. 5), you will reduce the overall tie trunk cost
without degrading service. With dial-tandem, you need only one hub;
a technique called "digit absorbing" allows you to have a direct
trunk group from one PBX to another, with an alternate route via the
hub. With CCSA, you cannot have PBX to PBX automatic alternate
routing unless you have very smart PBXs. Within CCSA, you must have
a minimum of three hubs before you can have any alternate routing at
all.
Traffic Information
CCSA has Automatic Message Accounting
associated with it. This equipment can be useful because it provides
you with point-to-point traffic information. You can tell what
proportion of the calls from each location go to every other
location. You cannot, however, tell the total number of calls
handled because a sampling technique is used: one call in five. It
turns out that the sample is up to one call in five, and may
be appreciably less.
Transmission Quality
One might suppose that CCSA would provide
better transmission than dial-tandem. This is not necessarily true.
CCSA was invented to consume No. 5 Crossbar systems when it appeared
reasonable that No. 1 ESS could take over the local switching
market. No. 5 is a fine machine, but it was designed to connect
local residential customers to trunks and vice versa. As such, it is
a 2-wire switch and is no more set up for trunk-to-trunk connections
than a 770 PBX. It can be arranged to do trunk-to-trunk connections
in CCSA (and it performs similar service in the public network) but
the process leaves much to be desired. And even when the connection
is established, it is 2-wire and has transmission properties no
better than would be obtained from SXS selectors. There are some
4-wire switches in CCSA, but don't count on having one. When No. 4
ESS is more generally available, this situation may change
considerably.
Alternatives to CCSA
It happens that many electronic PBXs must, of
necessity, be 4-wire internally. This indicates that tie trunk
switching can easily be implemented in an electronic PBX, solving at
one thrust the 4-wire switching problem and eliminating the
access-line problem as deftly as could be done with a 701. A
separate switch for tie trunks at or near a major company location
would seem most undesirable.
With one switch serving the needs of a single
customer, special programming for alternate routing, off-net access,
etc., might be more easily obtained than in a switch shared with
other users. Then, too, the use of a single switch handling local
extensions as well as tie trunks and providing SMDR for both offers
some real possibilities for savings. In any event, elimination of
access line charges alone might go a long way toward justifying the
cost of such a machine.
This is not the entire story, however, when a
major PBX is located at a considerable distance from a high density
point, or if tie trunks are obtained from a specialized common
carrier that also offers switched service. In either instance, the
extra cost of bringing all external tie trunks to the remote PBX for
combined switching vs. adding a somewhat smaller group of access
lines to a centralized switch must be traded off. A low-density
point increases the cost of AT&T circuits, and the cost of extending
specialized common carrier circuits from their major terminating
locations to customer premises via telco end-links must be added to
the cost of the specialized circuits themselves. Thus, depending on
the location of the customer's PBX, the use of internal vs. external
tie trunk switching is not always obvious. Accurate traffic
information is required to size the access group properly. Only when
you know how many calls originate and terminate at the hub-PBX and
how many are switched through can you make an accurate cost
evaluation.
When considering a CCSA-type switch from a
specialized common carrier, there are two other points to be
considered. First, will the specialized carrier be able and/or
willing to interconnect access lines, or will it only connect access
lines to its long-haul tie trunks? If you can't call from one
location to another through the hub, you may not want the service at
any price. Second, will the specialized carrier CCSA hub permit you
to use its routing and charging capabilities to access WATS lines?
If you can't use your nearby CCSA switch to provide centralized WATS
access for many small locations, each of which could not justify
WATS on its own, again you may want to forget the whole thing.
THE BROADER ALTERNATIVES
By and large, tie trunks are used during the
business day and lie idle after five o'clock. I have heard estimates
suggesting that at least a third of all intercity circuits are
private tie trunks (although not all are in switched voice
networks). If this largely unused capacity could be accessed by
residential callers, most of whom make their long distance calls
after five and on weekends, the obvious efficiency of the
arrangement would warm the cockles of any engineer's heart.
One would suppose this would be the actual
motivation behind CCSA: to put tie trunks in central offices where
they would be accessible to residential users as well as business
users. This does not seem to be the case, however. The chance to
offer "business day" tie trunks to business users at a discount rate
and to make the same trunks available to all after five PM appears
to have been lost. But it would have been a good idea.
The real need, of course, is to switch tie
trunks on a 4-wire basis, and that is hard to do with either the
crossbar or reed-switch equipment typical of local central offices.
Business customers, with non-voice as well as voice traffic, could
well use 4-wire switching, not only for tie trunks but also for
local connections (as, for example, from a terminal to a computer).
The obvious approach is to use digital central offices, which are
4-wire internally and can access remote line concentrators on a full
4-wire basis. These concentrators could, of course, be PBXs
rendering Centrex service.
With the option of local or remote PBX
switching units, required by the particular installation, a central
office switch for both residential and business customers could
provide Centrex service combined with CCSA to eliminate access lines
between the two, make tie trunks available to the residential users
after hours, and to generally lower costs and provide better
utilization of facilities. The Bell System, however, seems to have
very little interest in digital switching for local COs; and the
Independent Telephone Companies, while moving in this direction,
serve primarily rural areas with reduced need for business
communication. Thus, we may never be able to take full advantage of
what the technology can offer along these lines. Perhaps it's just
as well.
References
Shea, P.D., Centrex Service: A
New Design for Customer Group Telephone Service in the Modern
Business Community. AIEE Communications and Electronics,
Nov., 1961, pp 474-482.
Spiro, George, Centrex Service
with No. 5 Crossbar. Bell Laboratories Record, Oct., 1962, pp
327-331.
Liss, W.A., and others,
Centrex Service in No. 1 ESS. Bell Laboratories Record, Nov.,
1968, pp 332-338.
Kraus, C.R., CCE Newsletter,
January, 1977.
***
I later found out that
CCSA's inability to go off-net via a distant PBX was just Bell
policy for some customers. Other customers had been provided with
the feature. The No. 5 Crossbar switches were arranged to seize a
trunk to the PBX, insert a 9 and a pause, and then outpulse the
seven digit number into the public network. All I can say is they
refused to sell it to me.
It doesn't matter much,
however. The Bell System apparently changed its mind even while I
was writing the article. EPSCS (Enhanced Private Switched
Communication Service) was replacing CCSA for tie-trunks, using
4-wire connections on a single pair through a No. 1 ESS tricked up
with fiendishly clever electronic trunk circuits; EPSCS is also
arranged to provide so much management information that the user
often can't find what he needs in the pile.
Perhaps more important,
new tariffs were and are being filed under the name of ESSX, where
Centrex type service is no longer considered suitable for large
customers or those more than a mile from the CO, but is dandy for
small customers near the telco building. This has caused problems;
faithful telco sales people, conditioned for years to say "Centrex
good, PBX bad" are now in a state of shock and confusion as they
have to say "Dimension good, Centrex bad." Looking to the future,
one wonders what will happen when the Bell System finally gets a
digital PBX to replace the Dimension. At present, many Bell System
sources are saying "Analog good, Digital bad." How would Drs. Pavlov
and Skinner deal with this problem in behavior modification?
[
Top ] [ Next ] [
Table of Contents
] |
