Voice
Communication in Business Volume 2
Essays on telecommunications,
1981-2002
Centrex
was always misunderstood. Thus I tried to clarify matters with this
paper in the May-June, 1986, issue of Business Communications
Review.
Ten Common Misconceptions
About Centrex
(Business Communications Review, 1986)
With the coming of
interconnect, divestiture and competition, Centrex was supposed to
fade away. It didn't. There is more of it now than two years ago,
and, based on the hoopla one reads in telecom journals, it is going
to keep right on growing.
Centrex has some real
advantages, which we'll get to eventually, but there are a number of
very serious misconceptions about Centrex floating around that may
lead the unwary astray. Under the assumption, possibly silly, that
people want to base major decision on facts, let's take a look at
reality, such as it is, and examine what I consider the ten major
misconceptions about Centrex, along with its four big advantages.
Misconception 1. Centrex was originally offered to provide business
customers with new features which they were demanding.
Nothing could be further
from the truth. Centrex was invented by New York Telephone to
minimize the cost of installing SXS PBXs, removing them after an
average of something less than three years, stockpiling them, and
then reinstalling them somewhere else. NYT reasoned that a modern
office building could be wired up, once and for all, to a switch in
the same or an adjacent building and partitioned to serve a number
of clients as needed. They rammed the idea down AT&T and Bell Labs'
throats, and Centrex came into being. Although the first Centrex
installation is generally cited as the Air Force Academy in Colorado
in 1958 (Ref.1), the main thrust came in Manhattan in the early 60s
in response to the post-war building boom in office space (Ref. 2).
It should be noted that some early Xbar Centrex systems (as late as
1974) could not transfer calls and do other things that PBXs could
do easily, and that the features of No. 5 Crossbar and, much later,
of ESS, that made these suitable for Centrex service were features
that served the telco, not the customer: namely multiple classes of
service and AMA (Automatic Message Accounting).
Misconception 2. When Centrex was introduced, PBXs could not provide
DID. Crossbar and ESS technology were required to implement Centrex.
As a matter of fact,
about half the Centrex systems installed up to, say, 1975 were
Centrex CU, based on 701 SXS PBXs rather than Centrex CO based on
Xbar or ESS switches. Any PBX that could handle incoming "dial
repeating" tie trunks could handle Direct Inward Dialing.
Unfortunately, most Xbar and early ESS Central Offices, due to the
almost incredible inflexibility of their architectures, could not
outpulse on the "line side" toward the customer. Thus CO limitations
often prevented PBXs, otherwise capable of doing the job, from
providing the customer with Direct Inward Dialing..
It is interesting to
note that the third DID installation in the US was at Dupont in
Wilmington, Delaware, approximately 1958. A SXS Central Office
simply put some of is final selectors on customer property, saving
90% on the number of pairs that had to be run.
Misconception 3. SMDR on Centrex gave the customer better control of
his calling.
Centrex added two
features to PBX service: DID, as we have seen, and Identified
Outward Dialing, or IOD. This service, long available on PBXs,
predates Centrex by perhaps a decade. Called "QZ Billing," it was
implemented by having a telco operator bridge the connection, obtain
the calling PBX extension number verbally, and enter it on the telco
billing record. With AMA or automatic message accounting, Xbar was
able to do such billing automatically, as was the later ESS; thus
the telephone companies could swap technology for people and save
themselves some money in the rendering of an old service.
Unfortunately, AMA, IOD
or A (for Automatic) IOD was of little value to most business
customers. It did not work on tie-trunk calls, FX calls, WATS calls,
etc. This led to the rapid development in the mid-70s of
customer-owned toll recording equipment which worked on all outgoing
calls. But customer-owned billing equipment worked better with PBXs,
and was ultimately incorporated in PBX designs after 1975. Because
it got all calls and not just those that went by the most expensive
route choice, PBX SMDR (or Call Detail Recording, CDR) became a
major reason to choose a PBX rather than Centrex.
Even those who used the
Centrex billing, however, found that it often provided less control
rather than more. Because the toll billing was by individual
extension, overall calling patterns of the company as a whole were
almost impossible to see, and, when computer analysis became
practical, added an extra step in the processing of AT&T tapes.
Misconception 4. PBXs could not provide Touch-Tone calling. Only
Centrex Could.
There is an element of
truth in this criticism of PBXs vs. early Centrex. Touch-Tone
detectors, made of individual transistors, resistors, capacitors,
etc., were expensive. Thus, their use had to be prorated over very
large numbers of customers to make the cost per phone competitive.
This worked better in large COs than in small PBXs. (Note that a big
PBX might have 1000 lines, while "small" in a CO would usually start
at 5000 lines and work up.) However, large SXS PBXs were available
with special signaling access to Touch-Tone detectors, incorporating
them easily, although at a cost. When electronic device fabrication
advanced enough to bring costs down, about 1975 or so, Touch-Tone (DTMF)
became common. On customer-owned PBXs, DTMF was usually more
economical due to phone purchase rather than rental at premium
rates. And even today, there are some COs that do not have DTMF or
Touch Tone.
Most of the other
features that were triumphantly announced for Centrex were no more
needed than in PBXs of the same period. Three-way calling was
retrogressive compared to appropriate use of 1A2 key systems, and
there was no way that Centrex Features, like those on newer PBXs,
could handle the secretarial screening that customers with 1A2 were
used to. Call forwarding in its various guises is something more
loved by designers than communication managers, and stand-alone PBX
restrictors did a vastly better job of cost control than Centrex
class of service. And of course PBXs could hunt, dial 9 for outside,
dial 0 for the switchboard attendant, etc. etc., before Centrex had
ever been thought of, and manual transfer, manual route selection
and manual toll billing remained highly cost-effective until about
1975.
Misconception 5. Centrex could save the customer considerable
amounts of money on floor-space.
This, like Number 4
above, has some truth to it. However, it was based on the assumption
by the telephone industry that floor-space costs, like telephone
costs, were quoted per month. It turns out that building managers,
at least in New York, quote floor space costs as so much per square
foot per YEAR. When disinflated by reality, floor-space savings were
thus often negligible. Indeed, they were sometimes negative.
Floor-space in the main office area for Centrex consoles, for one
client of mine, turned out to cost more than the cost of basement
space for a 701 SXS machine and a 5 position manual board. Further,
some electronic PBXs after 1975 turned out to be smaller than the
cross-connect frames needed to take all the telephone wires back to
the Centrex switch. Today, floor space savings have to be considered
very carefully if they are to give correct information in any
particular instance.
Misconception 6. Centrex is non-blocking.
Where this one came
from, I do not know, but it has turned up in several articles I have
read in the last three months. It is absurd. CO switches handle very
large number of lines and, because the complexity of a switch is
roughly proportional to the square of the number of lines served,
the switching matrix cost per line is higher in large systems than
small. Thus concentration is commonly used, CO switches are
engineered for a good grade of service, and are less likely to be
non-blocking than PBXs. Indeed, one writer, primarily knowledgeable
in the CO switching field, suggests that building a non-blocking CO
switch might well be illegal. (Ref. 3).
Actually, the only
reason to consider a non-blocking switch is to avoid the
administrative cost of "traffic balancing" over the life of the
switch. That is, one can assign heavy and light users of telephone
connections to any available ports on the switch without fear of
degrading service for others. Note that a non-blocking matrix does
NOT guarantee you can always get your connection. There is nothing a
non-blocking matrix can do for you if all trunks are busy, or if the
person you want to talk to is already on the phone. In high-traffic
systems, designers have to ask how much a blocking probability of
.0001 or less is worth if the probability of the port desired being
busy is .25 or more. Justifying such costs to a PUC, necessary in
Centrex but not in PBXs, is sometimes difficult.
Misconception 7. Because Centrex is in the telco Central Office, it
has maintenance personnel available at all times.
This looks reasonable on
the face of it, but it is not quite what might be expected. It turns
out that the reliability of ESS offices is quite high, and no one
office will develop enough failures to keep the trouble-shooting
skills of a maintenance team active. Thus several ESSs have to be
served from some central location to adequately exercise a
maintenance group.
Another point of more
than passing interest is the way, in current Centrex, station
equipment has to come from someone other than the regulated local
telephone company. This means that somebody has do decide whether
the fault is in the sets or in the telco equipment. With two
different groups responsible, "finger pointing" becomes a natural
result.
Finally, with station
wiring going over much greater distances, the exposure to trouble is
higher. PBX station wiring is short and usually within one building.
Centrex wiring is exposed to outside conditions including lightning,
power crosses, back-hoes, etc. There is even a greater opportunity
for wire-taps.
Misconception 8. Because Centrex is implemented with Central Office
equipment, the customer always has the most up-to-date facilities
available.
We all know this is
simply a marketing ploy. As we have seen, many early Centrex systems
could not transfer calls, and even today, many can only do SMDR on
certain kinds of calls. ARS and restriction are typically much less
satisfactory in CO switches than in good modern PBXs, and the
biggest barrier to implementing the ISDN in the United States is the
vast deployment of relatively new ESS switches with space-division
matrices. These 2-wire analog switches are just fine for the
residential service for which they were designed, but they are at a
real disadvantage in providing service for large business customers
who have extensive needs for nonvoice communication (that is, for
Centrex customers).
Misconception 9. Centrex is the ideal vehicle for handling data.
Centrex, like any
switching system that can deal with voice frequency bandwidth, can
handle modemized data fairly well. Further, because there is no
particular limitation on the bandwidth that reed switches can pass,
an ESS should be able to handle data bandwidth in the 10 to 100 Khz
range if necessary (it has even been checked out for Picturephone
bandwidths). Unfortunately, the long loops from CO to customer
premises, typically much longer than PBX station wiring, limit such
broadband transmissions. Practical frequency range is limited by the
electrical "capacitance" between wires which is directly
proportional to the loop length, inversely proportional to the
spacing between the wires, and also related to the kind of
insulation used. Further, long loops increase exposure to electrical
noise. Thus the potential broad bandwidth of reed switches is
limited by the wires needed to reach them.
Digital PBXs deal fairly
well with all these problems. A digital PBX has to be four-wire;
that is, it has a separate path from me to you and from you to me.
With two separate paths, high speed full duplex data transmission is
much easier than in a two-wire system. Second, because PBX loops are
short, noise pick-up is minimized. Further, the cost of copper is
relatively unimportant in PBX wiring compared to the cost of labor
for installation. Thus two or more pairs can be used for station
wiring without the cost penalty that would exist in doubling the
much longer CO wiring, and transmission to the telephone can be done
on a four-wire basis. Finally, the codec, used on a per-line basis
today, can be put in the set as easily as on the line-card. Thus
digital transmission, with its inherent noise immunity, is available
today within a modern digital PBX. Because voice coding commonly
uses 64 Kb/s, rather high speed data can be moved as though it were
a voice signal, and voice and data can be multiplexed together on
the same wiring. Further, the same physical port on the matrix can
be multiplexed into two or more separately addressable ports and
voice and data can be switched independently.
The way Centrex systems
are handling data today is relatively ingenious. They are putting in
separate equipment between the analog switch and the loop to the
Centrex telephone. Then, on the far end, they are putting a separate
interface box between the loop and the voice and data terminals. The
two boxes, one in the CO and one at the work station, make a
separate path for data, and bring it out for separate switching.
Some such systems use frequency division, deriving two channels, one
to carry data in each direction, at frequencies well above the voice
band. Others code voice and data alike (as in a modern PBX) and use
time division on one pair for the two directions of transmission,
ping-pong fashion (bit compression multiplexing). At the CO, voice
is returned to analog for conventional switching, and data is passed
to a separate data switch. Another technique takes the combined
voice and data signal and transmits in each direction continuously.
At each end, the digital transmit is separated from the receive by
using hybrids. Where some signal still leaks through, echo
cancellers can be employed.
The box at the central
office can be a whole switching system in itself. Two typical
devices are DACS, or Digital Access and Cross Connect System, and
machines like AT&T's Data Kit. DACS is an electronic crossconnect
panel with the same functions as a Main Distributing Frame: it can
patch circuits through on a pegged-up circuit-switched basis. One
major advantage of DACS is that it can be controlled remotely. Voice
signals go to T-Carrier type channel banks which make the A/D
conversion to the analog switch (a direct connection is, of course,
possible with a digital CO) while data is routed elsewhere. DACS can
normally patch full 64 Kb/s voice channels as required, but
recently, AT&T (Ref. 4) has described a sub-rate patching capability
for data at lower speeds such as 4.8 Kb/s as well.
A Data Kit is an AT&T
packet switch for data. It establishes a "virtual circuit" between
two addresses but, like most packet systems, only uses the switch
bandwidth when a small block of information is present. Then, the
entire bandwidth is used very briefly to move the information from
input to output. Transmission channels between Data Kits can also be
used, so that packets can go between individual Data Kits on their
way from one user to another.
With this sort of
equipment added to an analog ESS in the CO, it is possible to define
Centrex service to cover both voice and data switching and
transmission. Whether it turns out to be cost effective or even
useful remains to be seen.
Misconception 10. Centrex will be ready for ISDN before PBXs.
Many digital PBXs today
have a direct interface to a T-span. If a similar connection is made
to a digital switch in the CO, we have a good start to an IDN if not
an ISDN. Because most toll switches, whether from AT&T or other
carriers, are digital while local CO switches, in general, are not,
and because fiber optics transmission is widely available between
toll switches today, long haul switched digital connections, whether
voice or data, appear to be possible right now. Such connections are
T1-D3 and not 23B+D as required by ISDN, but when 23B+D can be used,
only a relatively modest equipment change will be required.
Centrex, on the other
hand, will have to wait until digital central offices such as 5ESS,
DMS 100, No. 5 EAX, etc., are in place before real ISDN operations
can begin. Even where such switches are in use, however, they are
serving customers largely on a 2-wire loop basis, and are meeting
toll networks with T1-D3 trunk cards just like those available for
PBXs. They can't really do combined voice and data until chips for
the ISDN 2B+D local loop are available (it would be silly to develop
digital voice-data to the telephone set on some other basis with
2B+D known to be in the works), and when they do, they will have to
depend on telephone sets developed by other people. PBXs have their
own proprietary sets for voice and data today, and many are in use.
Internal to the PBX, they will continue work fine, independent of
2B+D, while for outside connections, their data signals can be
mapped into T1-D3 now and 23B+D in the future with about equal ease.
Thus Centrex does not
seem to be able to offer combined voice-data as easily, now or in
the future, as many of the digital PBXs on the market today. And
PBXs do not need to ever consider 2B+D, switch to set, if they don't
want to, while Centrex must wait. The delay may be too long for
many.
Now that we have
disposed of these misconceptions about Centrex, what does Centrex
actually offer the customer? As nearly as I can tell, there are four
big advantages that, in many instances, will overpower the way PBXs
can do voice, data, or anything else.
Advantage 1. Central Office battery.
Telephone central office
switches have a large battery to maintain their operation over
momentary failures in commercial power. On top of that, most COs
have a diesel-driven generator and a 30 day supply of fuel. Thus
thirty days after the world comes to and end, you can still get dial
tone. If you are a hospital, police station or similar operation
where telephone service is vital, particularly when the lights go
out, there is no better way to go than Centrex.
Advantage 2. You can Centralize multi-location organizations.
When a customer has many
nearby locations, Centrex is one of the best ways known to provide
one telephone system for the whole lash-up. City governments find
this particularly useful: City Hall is here, the Police Dept. is
there, the Fire Department is somewhere else, as is the Board of
Education, the several school buildings, the library, etc. Yet all
are close enough to be served by one central office as part of one
big "PBX."
Banks, too, find Centrex
useful. In many cities, banks have a branch in almost every block,
and other banks at greater distances. By using Centrex nearby and
OPXs when necessary, they can put all their branches on one system
and take advantage of security procedures that only permit Centrex
extensions to access the computer (for instance).
With common channel
signaling between control computers, "city-wide Centrex" can be
implemented, using several CO switches in the same general area to
handle company locations near them. The data link from one CO switch
to another need not be CCITT # 7 signaling; it could use earlier
versions available now such as the CCIS signaling in the AT&T long
distance network.
Advantage 3. With Centrex centralization, you can provide
centralized access to long distance facilities.
"Same or adjacent
exchange" tie trunks are going up rapidly in price, as are mileages
for OPXs and FX lines. However, almost identical facilities used for
Centrex stations are priced much lower. Thus a PBX is at a real
disadvantage in trying to build up toll usage by combining calls
from several nearby locations for volume discounts; the cost of
access is more than the possible savings. But when Centrex switching
is the same equipment for several locations, and has direct access
to long distance carriers, you just can't lose.
Advantage 4. With Centrex, you don't have to buy; you can still
rent.
This, in my opinion, is
the single biggest advantage of Centrex, and the principal reason
why Centrex is doing all right in a highly competitive market. Even
though the FCC has decided that customers should be denied the
freedom to rent if they want to, local PUCs are not giving
Washington much help. Since freedom of choice is the name of the
game, even when it conflicts with the DC party line, this has to be
a good thing.
But when would you
rather rent than buy, particularly when you have to buy your
telephone sets anyhow? There are lots of instances. The classic is
when you want to use your capital for our own business, not the
telephone business. Raising money is hard enough, without having to
tie it up in complicated machinery that needs, by its nature, a long
time for payoff. Another instance where renting may be better than
buying is when you only have two more years to go on your lease
before you move. Delaying purchase can save you, at the very least,
a second installation cost and, in the meantime, Centrex size can be
economically adjusted, either up or down.
Finally, you may prefer
to rent from a telephone company simply to eliminate maintenance
problems. With owned PBXs, maintenance contracts are getting more
and more expensive, and hiring your own technicians has special
problems. You can rent Centrex "just like in the old days," and
expect help from the state PUC if things don't work. You can even
refuse to pay until the system functions, something that you can't
do with an owned PBX, particularly when some financial organization
with no responsibility for system operation owns the paper.
Yes, renting is
sometimes a good idea, but an even better idea is to insure the
customer's freedom of choice to rent if he wants to. That, alone,
may be the single most important advantage of Centrex in American
telecommunications today.
REFERENCES:
1. Spiro:
Centrex Service With No. 5 Crossbar. Bell Labs Record,
October, 1962.
2. Shea:
Centrex Service. Communications and Electronics (AIEE),
November, 1961.
3. Briley:
Telephone Switching. Addison Wesley, 1983.
4. Gerrish and
Morrison: The DACS Door to the Subrate World. Record, AT&T
Bell Laboratories, January, 1986.
[
Top ] [ Next ] [
Table of Contents
] |
