Voice
Communication in Business Volume 1
Essays on telecommunications,
1969-1980
Chapter 13
Restrictors, Recorders & Routers: Built-In Or Add On?
Most
PBXs, when they first came on the market, did some of the things
that PBXs had always done, but very few considered the need for
automatic route selection, detailed billing by extension, etc.
Observing these and other omissions, it is both sad and funny to
reread articles in the technical press where designers lamented the
lack of sufficient customer appreciation of their new hardware. The
very kinds of things that computers could do best were, all too
often; not even considered by these "innovative" designers.
There
were others, however, who looked at the problem and came up with a
different approach. These people saw that add-on systems for use
with existing PBXs and other telco-supplied equipment could provide
the customer with telephone company reliability for communication
service, and interconnected ingenuity for services not previously
available. Stand-alone restrictors came first, and long before
interconnect; LSI made it possible to redo the job. But recording
systems, routing systems and processing systems using either
recorders or routers as information sources sprang into existence.
It would
be nice to say this sort of competition has caused a lot of
interactive innovation, and to some extent it has. However, the many
design groups involved have gone off in various directions, seldom
monitoring the work of others and all thinking they are leading the
parade. There are many good ideas out there, embodied in hardware
and software, but as with PBXs, one seldom finds one system with all
the good things and none of the bad.
While
assembling the various chapters of this book, I discovered that I
had written very little about add-on equipment. Because I had had
several interesting experiences with add-ons, some quite spectacular
but others differing markedly from the more enthusiastic claims of
their vendors, I rolled out the word processor to fill the gap. The
result below first appeared in BCR for November/December, 1980.
***
Perhaps the most
discouraging thing about the telephone company in "the old days" was
the way its salesmen told you that you didn't want what you said you
wanted. And nowhere were they more adamant than when dealing with
customers who wanted information for cost allocation, particularly
on tie-trunk or WATS calls. "Why don't you treat your WATS lines and
tie trunks like electricity or air conditioning?" they would say.
"Just divide the bill up by number of telephones and be done with
it."
It did little good to
explain that some departments lived on the telephone, while others
seldom used it at all. And if you suggested that personal calls were
being made on company lines, and even on WATS lines in violation of
the tariff, they just smiled.
In all fairness, it must
be remembered that the need for automatic toll recording is
relatively recent. Not too long ago, long distance calls were much
less common than they are today, and there were no computers to sort
and digest usage data. Manual recording by PBX extension was cost
effective, and even the telephone company offered it on toll calls
as QZ billing. When large companies started getting their own
computers, their switchboard attendants used "mark sense" punched
cards to make out toll tickets to reduce the cost of automatic
processing. Some companies, as a matter of fact, have only recently
discontinued this approach.
Starting about the time
of Carterphone, which also matches the start of relatively
inexpensive solid-state logic and small general purpose computers
for use as part of specialized equipment, we find a number of
machines coming on the market to help the communication manager
control the usage costs generated by his company. Today, a decade
later, such machinery has increased greatly in sophistication, and
can be added to any PBX, old or new. But PBXs have increased their
abilities too, based on the same hardware. Thus a choice between
add-on and built-in equipment is often required.
In either case, any
approach to cost control involves two parts: permitting only the
right calls, and making those calls go on the right facilities. With
regard to the right calls, restriction in all its various forms can
prevent the wrong calls from being made before the fact, and call
data recording can make users accountable after the fact.* Choosing
the right facility involves route selection, either manual or
automatic, and sometimes queuing to obtain better utilization of
those circuits where higher use results in lower cost per minute.
Add-on and built-in systems have very different properties in these
several areas, and care must be taken in their selection.
[* FOOTNOTE: It is desirable to shake
personal calls out of the system before buying an expensive
standalone, either router or recorder; pure business traffic may not
be enough to Justify the investment. Sometimes equipment can be
rented, but often processing the toll tape by called number is very
revealing.]
Restriction and toll diversion
Almost from the very
beginning, automatic PBXs provided some form of restriction, and
central offices provided toll diversion. Built-in restriction
usually worked only on the first digit dialed; an unrestricted
station user could dial 9 and get out, while a restricted user would
get reorder tone or be connected to the attendant. Toll diversion,
built into the CO, simply class-marked the PBX trunks for local
calling only, and caused reorder tone to be returned when a toll
call was dialed. As a variation, it was sometimes possible to get
"coin box service" which permitted a somewhat different dialing
range, even when coin control was not employed.
Both North Electric and
Stromberg-Carlson have produced elaborate restrictors for the past
30 years. Restrictors allow toll calls to certain areas while
blocking them from others. They intercept CO trunks as shown in Fig.
1, and originally looked at the first three digits dialed (usually
the office code) to decide whether or not to pass the call. With
DDD, it became necessary to look at the first six digits, to permit
screening on both area and office code. Prior to the coming of
modern electronic memories, a result of large scale integration,
this was quite a task. When read only memory (ROM) appeared, the
previously difficult job became simple, The Phonetele restrictor
makes excellent use of this new technology.
A restrictor does not
know which station originates a call. It tests all calls alike as
they pass by. Thus restriction is a function of a particular trunk
group, and is not based on the "class mark" of the calling party.
But this is a blessing in disguise; even the boss cannot use the
wrong trunk to place a call, and restriction is seen to be the poor
man's automatic route selection. By only letting calls go through if
the user has dialed the access code to reach the right trunk group,
FX lines, WATS lines, etc., can be loaded effectively.
Modern restrictors can
block abuse and force calls into the right trunks; thus, any PBX of
any vintage can be given this advantage at small extra cost. But
restrictors do not queue calls, and their whole approach forces the
user to select the proper trunk group. Testing one group after
another for a free circuit, as in route advance or more complex
automatic route selection schemes, is thus not practical. But where
they work, they do an excellent job.
PBXs have slowly
increased their restriction capability. With the appearance of DDD,
area codes had to be dialed; the second digit was always a 0 or a 1.
Further, within some areas, local calls could be dialed with a 7
digit number, while toll calls required a 1 or 0 to be dialed first,
the 0 picking up an operator for some sort of assistance. By looking
for a 1 or a 0 in the first or second digit dialed into the public
network, it was possible to build restriction into PBXs very simply.
This form of
restriction, although easy to implement (even before the coming of
computers), may not be useful much longer. Office codes are
beginning to be assigned with 0 and 1 as their second digits, just
like area codes. This adds 20% more office codes to those that are
possible within an area, but it makes the second dialed digit
unsuitable for detecting toll calls in some areas. An unfortunate
consequence is that now the user must know when to dial 1 or not as
a first digit for a call within his home area code. For the computer
to add or omit the 1 is much more difficult than when a simple
look-up table could be used.
Many PBXs today
incorporate six-digit screening. Their biggest advantage over
external restrictors is their knowledge of who placed the call. They
can read the class of service stored in the processor for the
extension making the call and decide if, for the call in question,
the caller is restricted or not. By giving different extensions
different calling ranges identified by different class marks, good
control can be exercised. For a stand-alone restrictor to do this,
identity of the calling extension would have to be passed forward
somehow along with the dialed digits.
Many enthusiastic but
naive planners will assign class marks with great care, giving
different class marks to phones at adjacent desks. What often
happens is amusing: a person, unable to make a particular call,
simply walks from desk to desk until he finds a phone that works.
Restriction on a class-mark basis is a very important feature, but
it must be applied with reason.
Recording, AIOD and AMA
From the beginning,
telephone companies have sold a service, not a product. You don't
buy a telephone, you buy telephone calls which the telephone, along
with a lot of other equipment, makes possible. Telephone calls are
charged individually, and the telephone company has to keep track of
each and every one to give you a fair and accurate bill.
Local calls don't cost
much, so very little can be spent for the equipment to figure the
charge to the caller. Thus, flat rate service has been common in the
past. You paid so much per month, and got to make as many calls as
you liked as long as they were in your "exchange area." For
residential phones, where the calling rate was low, and in small
cities where one central office covered the whole area, this was not
a bad deal. However, when a business made many calls, or when many
central offices were needed to serve a local area, as in New York or
other major cities, it was necessary to adopt "usage sensitive
pricing." Message Units were invented, and the caller was charged
one or more message units for each call.
Message registers were
installed on each line, and each call was counted. Calls going a
longer distance within the local area might hit the message register
after each unit of time; sometimes two or more counts would be made
for each interval, depending on distance; calls within the same
central office would often be charged one message unit only,
independent of time.
Message registers had to
be read each month, and the message unit totals found. To reduce the
manual labor involved, a photographic plate picturing a large block
of registers would be exposed twice, once at the beginning of the
month and, a little higher, at the end of the month. With the
picture of message registers developed, the end of the month reading
would be just a little above the beginning of the month; it was thus
easy to subtract to find the number of message units used. I
actually saw this method in use in the mid '50s.
Toll calls were
originally ticketed manually. To permit Direct Distance Dialing
(DDD) by customers, it was necessary to invent Automatic Message
Accounting or AMA. AMA was put into service long before computers
were available, and Amos Joel of Bell Labs has what may still be the
largest patent ever issued to cover the system. Early AMA was "three
entry." The first identified the calling line, stored the called
number, and recorded some other information needed for data
reduction. Recording was associated with the outgoing
toll-connecting trunk; thus the trunk number was stored. The second
entry was identified by the trunk number and recorded answer time.
The third entry again used trunk number for identity, and recorded
hang-up time. Note that this approach recorded events as they
happened, with calls interleaved in time. However, it eliminated the
need to store large quantities of information temporarily in relays
or other large, expensive devices. Some early electronic toll
recorders blindly copied this three-entry technique, even though
newer memories had long since offered better approaches.
AMA required the
identity of the calling line. For residential service, and even
small businesses, this was fine. But billing all PBX calls to the
PBX directory number left the communication manager with a problem:
he could not charge calls back to the extension that placed them. In
some instances, he might use separate CO trunk groups for different
departments to get a telephone bill for each department, but that
was about as far as he could go. For billing by extension, the
extension number had to be made available to the CO that housed the
AMA equipment. Very small central offices, known as Community Dial
Offices or CDOs, had a similar problem. QZ billing, or something
quite similar, had to be employed: an operator would be connected to
a toll call; after obtaining the called number, the operator would
then key it into the AMA equipment.
To do the job
automatically, "identifiers" were invented. When the CO discovered a
toll call had been placed, it would call the PBX or CDO via a data
link and ask for an extension identity. The identifier would find
the calling extension number and transmit it to the CO for addition
to the AMA tape record. Automatic Identified Outward Dialing, or
AIOD, is widely used today; modern PBXs include the ability to
identify calling lines and pass the information forward. But note
that AIOD only works on toll calls; it does not work on Message Unit
calls, WATS calls or calls via tie trunks.
"AIOD" identifies the
process where a PBX extension's identity is passed forward to the
local CO to complete the telco toll record. Many people use the term
incorrectly to refer to any system that provides a detailed bill for
toll calls generated by each extension.
AIOD, along with DID or
direct inward dialing, became the major items that, added to PBX
service, made up Centrex. Centrex CU, however, did not need to
provide automatic IOD. Operator identification, using former QZ
operators, was common. Centrex CO, of course, used Central Office
equipment rather than a modified PBX on the customer's premises; the
CO equipment usually included AMA as a built in feature.
Because of the
limitations of QZ service and AIOD, the revolution in solid state
components quickly produced tape recorders to store similar
information at the PBX. Such "stand alone" recorders bridge each
extension line and record all its signaling activity as shown in
Fig. 2. With dial pulsing, this was quite easy and inexpensive. Even
with several hundred inputs, there was no big problem; just a lot of
wire. With the coming of DTMF or Touch Tone Signaling, special digit
receivers had to be associated with each line while keying was in
progress. This ability, added to the bridged connection to each
line, made the recording systems more complex.
Stand alone recorders,
of which those made by Alston are typical, can be added to any of
the older PBXs or, for that matter, central offices. Message
register photography has been replaced in recent years in many
elderly COs in just this way. The big advantage that stand-alone
recorders offer, particularly to business communication managers, is
complete knowledge of station behavior. All calls are recorded:
intra-PBX, local, long distance, FX, WATS, tie-trunk, etc. Further,
aborted and incomplete calls are noted; this kind of information
provides a picture of user retries, and also helps identify bad
trunks, tie trunks and other facilities. Often the information about
unsuccessful calls is more important than the scoop on calls
completed.
Some of the newest PBXs,
including the SL-1 and Rolm with Electronic Key Telephone Sets, send
their signaling information in digital form. A dial pulse or DTMF
recorder cannot pick up such signals, and a stand-alone recorder
cannot be used. However, these newer PBXs now offer better ways of
handling the problem.
Most modern PBXs have
been modified since their introduction to facilitate the recording
of call details. Call Detail Recording (CDR) or Station Message
Detail Recording (SMDR) is now generally accepted as a basic PBX
function. However, few PBXs include the recording mechanism itself.
Although most modern PBXs create and buffer the call record
internally, they must have a place to put call records when their
internal buffer is full. Some of the smaller feed information to a
"current loop" that will run a teletypewriter or similar printer.
The communication manager ends up with a list of calls in
approximate chronological order. For less than 20 feet of TTY paper,
this is often acceptable. Some TTYs can also punch paper tape,
making a machine-readable record that can be processed by a
computer. For a large number of calls, machine processing is
absolutely necessary.
Paper tape is not the
best input mechanism, and magnetic tape is preferred. A regular
9-track recorder of the type used in computer installations is quite
expensive, however, and can usually be justified only in very large
systems. Information is not sent via a current loop but, rather, via
an RS 232C data interface. With this standard interface, any sort of
data recorder can be attached as shown in Fig. 3. Digital cassettes
are often used in smaller systems, and are quite satisfactory in
most cases. At the moment, the floppy disk is coming into high
favor. A floppy disk system can easily be added to or built into a
PBX in very much the same way a cassette tape drive can, but it
operates faster, being more or less random rather than serial
access. Floppies are particularly good for storing back-up programs
or station parameters, and can reload the system quite rapidly after
a power failure. When handling the SMDR data, they make it easier
for the communication manager to change the recording medium when it
is full.
Recording call details
is only half the problem, however. Once you find yourself clutching
a roll of tape or a stack of disks, you realize that the recorded
information must be processed to produce meaningful results. You
need various sorts to put the data into useful categories, and
various additions and manipulations to get extension bills and
management reports.
Some add-on tape and
disk systems permit remote polling; sometimes this capability is
provided via the PBX. In either approach, a remote processing
company can call up your system and extract the recorded data for
processing. A few days later, a complete set of reports is delivered
to your desk. This makes it unnecessary for the communication
manager to remove one tape, replace it with another, deliver the
tape to the processing center, etc. Use of an external center is
not, of course, required. However, people who process telephone
tapes for a living have usually evolved some rather sophisticated
programs that even a good programmer could not duplicate without
extensive experience; the cost of developing these programs has been
prorated over many clients, and their use is usually quite
satisfactory.
Many service bureaus
that process tapes will also make their software available, for a
fee, for use on the client's computer. However, use of the corporate
computer for telephone work is often impractical; sometimes there is
simply no time available on the machine if the computer is being
used effectively in the service for which it was obtained.
To get around this
difficulty in obtaining highly specialized processing when needed,
several floppy disk systems that not only record but process call
details and print out reports have come on the market. Telephone
Management System's ZapCall and ComDev's CallQuest are two of the
better known. These machines simply plug into the RS-232C or current
loop output from the PBX, and absorb information just as a tape
recorder would. But the computer revolution allows them to sort the
information in a variety of ways and, upon request, print it out in
various forms.
These add-on
processor-recorders are powerful tools. They enable even a very
simple PBX to provide sophisticated management information when
desired, but they need not be added when the extra cost is not
justified. However, some of the most sophisticated PBXs available
today do nothing more than format and output individual call
records. Thus, the add-on recorder-processor appears to have a
bright future.
Call routing systems
Automatic call routing
systems, exemplified by the Action WATS-Box, the Datapoint
Infoswitch, and the WATS Commander from Commander Systems, take full
advantage of today's small computers. A computer with proper
peripheral equipment can accept or reject a call on the basis of
very complex rules (class marks), choose the right facility on the
basis of rules of similar complexity, and then record the
information needed to provide bills for user accountability and
system management. It is remarkable how few modern PBXs considered
this excellent utilization of modern technology when they first came
on the market, and how reluctantly they have added it in the face of
competition from stand-alone add-ons. Even today, the stand-alones
usually offer far more sophisticated features and services.
However, a built-in
routing-recording system has the same advantage that a built in
restriction system has: it knows, automatically, the line that has
originated a call and, hence, that line's class marks. The user does
not need to take any action at all to get this information to the
call routing mechanism. In addition, a PBX must handle all calls,
whether or not they are passed on to a stand-alone router; thus, it
can apply route selection on message unit and short haul toll calls
which, because of their very high volume and low cost, cannot
economically be handled by a stand-alone. Overflow to toll as a last
resort is more economical than with a stand-alone because no
expensive path through the stand-alone is tied up with a call that
cannot save money.
Next, it is often easier
for a PBX to provide callback queuing than it is for a stand-alone;
for small trunk groups, call-back queuing is far more effective for
packing calls into special facilities than off-hook (hold-on)
queuing. Finally, PBX designers have, of necessity, had to give some
consideration to the transmission properties of their switch. Sadly,
this cannot be said of the designers of most stand-alone systems.
There are certain circumstances where some stand-alones provide a
connection through which the callers can talk only with great
difficulty.
To see how the
approaches differ, let's trace a call through both types of system
and see what happens. With a built-in, the station user comes
off-hook and dials his call. Where the system is set up correctly,
the called number is dialed into the system memory and filed there
along with the calling extension number and the appropriate class
marks. For a permitted call, the system starts looking for an
outgoing facility as soon as the first six digits are dialed. If it
finds one, it seizes it and starts outpulsing. With the user still
dialing in, the PBX cannot outpulse any faster. Using this "overlap"
outpulsing approach, "overhead" or non-useful occupancy of the trunk
is increased. If no trunk is free, the PBX collects the rest of the
dialed digits and puts the call in queue. When a circuit comes free
or when time-out in the queue takes place, the system then completes
the call, if necessary via DDD, usually giving the caller a tone to
let him decide whether or not to try later on a less expensive
facility or go now at higher rates. When call-back queuing is used,
the PBX gets the calling party back on the line before it tries to
outpulse on the trunk it has selected.
With a stand-alone, the
user has to dial one access code for local calls and another for
long distance calls via the router as shown in Fig. 4 (this is often
easy: dial 9 for local calls and 8 for long distance). Then, upon
response from the router, he must send forward his account number or
extension, followed by the number called. The router can then check
class marks based on the identity number dialed in, and it can also
check the entire dialed number to block "trash calls" to off-track
betting, dial-a-joke and the like. Although screening by the entire
dialed number is also possible in PBXs, few designers consider it
worth doing. Study of frequently called numbers proves them wrong.
If the call is
permitted, the stand-alone finds an idle facility or puts the caller
in a queue; upon a facility coming free or time out, the call is
completed. Usually, the router will not even try to select a trunk
until dialing is complete; if the caller abandons the call before
dialing all the digits, no expensive trunk has been tied up
needlessly; further, when the system has all the required digits, it
can outpulse using DTMF in a very short period of time compared to
user dialing. Thus, once again, overhead time on the expensive
facility has been reduced.
If a call is to be taken
DDD after all other routes have been found busy and the queue has
timed out, the stand-alone router is at a disadvantage as has been
mentioned. An incoming and outgoing port are tied up on a call that
will not produce a saving; these ports add extra cost to the router
and the incoming port, while busy, tends to block other calls that
might be handled a little later on other facilities. If the router
could hand such calls back to the PBX for completion, the overall
operation would be more economical; unfortunately, nobody is set up
to do this.
Routing algorithms tend
to be relatively simple, but modern conditions require large amounts
of memory. In particular, the "specialized" common carriers only
cover certain down-town central offices in certain area codes. Thus,
a routing algorithm must look at both the office and area code to
see if a specialized carrier's network will take the call; with
about 125 area codes to study and up to several dozen central
offices in each which can be reached, the need for memory is fairly
large. Note, however, that one does not queue for the use of
"specialized" carrier switched services; these services are billed
by the call, and there is no advantage to queuing.
Queuing algorithms in
stand-alone systems and PBXs alike sometimes have a problem that
needs attention. Very frequently, the first choice route is an FX
line, with overflow to WATS. With several different FX lines
skimming the cream to major cities, building a queue is difficult.
Ideally, one would want to put each call simultaneously in a queue
for its FX line and also for the common overflow group, so that if
either an FX or WATS line comes free, the call will be served. Many
systems cannot do this; they only queue for the last group tested.
Thus, in our example, an FX line could come free but not be given to
a call waiting in the WATS queue. Queuing for two groups of
facilities in time sequence is also a problem, since a
first-choice FX line can come free and be snapped up by a new call
while an earlier call, now in the WATS queue, is still timing out.
From the user's point of
view, the main difference between add-on and built-in is the need to
dial an identity code of some sort, additional digits required by
the router for class marking and billing. Up to five digits per call
may be required; use of a repertory dialer is highly desirable.
However, note that the accuracy of these digits depends on the
honesty of the caller. It is quite possible to key in somebody
else's identity code and let him pay the bill. (This same problem
plagues customers of Execunet, City Call and Sprint.) On the other
hand, a PBX with a built-in routing system cannot protect against a
person going to somebody else's phone to place personal calls.
Use of the call-back
principle helps avoid this difficulty. If the system knows the
originating extension, and the user is trained to hang up after
"placing his order," the system will call him back, either
immediately or after a delay in queue. If the user dials the wrong
extension number, the system calls back that phone and the caller is
thwarted. Call back has its problems, too, however. If the caller is
busy on another call, the PBX may hunt to a secretary or be
call-forwarded to another office. It is a little unnerving for
someone to pick up the phone only to hear audible ringing (ringback
tone) on a call to an unknown person. Distinctive ringing on a PBX
helps in this situation, but a stand-alone router cannot invoke the
PBX "call-back" ring.
For stand-alone systems,
call-back permits a considerable economy in that a switching matrix
is not needed to connect the user-placed call to the appropriate
outgoing facility. The user dials in on any input path to the router
and then hangs up. The router, upon finding an appropriate outgoing
facility, connects it to an incoming path from the PBX (the same
kind used to place an order) with a relay or some such device, and
calls the user back through the switching matrix of the PBX.
Off-hook or hold-on
queuing has, as its main advantage, simplicity. The PBX or router
using off-hook queuing does not have to go looking for the calling
party when a trunk comes free—the calling party is right there,
waiting. With a speaker phone or a bridged amplifier as is found in
many modern telephone sets, delays can be longer than when the
caller must clutch a standard phone and wait; in either instance,
the phone is tied up during the queue as well as the call, and
cannot receive incoming calls. But the phone is ready for its
outgoing call with no further effort on the part of the router. Note
that inputs to a stand-alone router are occupied with calls in queue
as well as calls in progress. When all inputs are occupied, the
router cannot accept new calls and the number of calls in the queue
is limited. This can give a false idea of grade of service since the
router will not be able to include in its recorded information any
data on calls hitting all trunks busy at the PBX/router interface.
Note that an outgoing
call through a stand-alone router must pass through two trunks and
two extra trunk circuits before it leaves the area of the PBX. There
is the trunk between the PBX and the router, with a trunk circuit
(or the equivalent) on each end; then, there is a trunk from the
router to the carrier. This trunk has an interface at the carrier,
depending on the type of circuit involved. A PBX with built-in
routing would connect directly via one trunk with its trunk circuit
on one end and the carrier's line or trunk circuit on the other.
At this point,
transmission difficulties may arise. The problem is at its worst
when Centrex CO is used instead of a PBX. All the extensions go into
the CO. Then, a group of trunks must come back to the user's
premises to the router, sometimes a distance of several miles. After
passing through the router, this same distance or a greater one must
be traversed to the CO or office of the "specialized" common
carrier. Attenuation from the extra circuitry at the router as well
as the several miles of additional cable makes the signal much lower
in level. Because the makers of toll routers are, in general,
unfamiliar with such niceties as idle circuit terminations, the
carriers can add only limited amplification to compensate for the
added loss (an unterminated two-way amplifier can scream like a
public address system with the mike in front of the speaker, and for
exactly the same reasons). Further, when amplifiers are inserted
only during the talking state, their amplification is not
necessarily as represented on the label. Because of some basic facts
of electrical engineering, gain of a two-way amplifier may be more
or less, depending on the circuits to which it is connected. Indeed,
sometimes an amplifier can attenuate the signal rather than amplify
it.
As with passive
recorders, means must be provided to go from recorded information to
bills and management reports. And again, as with passives, there are
several options: some systems process their own data and provide
reports directly, others provide a tape or disk that can polled
remotely or sent to an external service bureau and, finally, the
tape or disk can be processed on the customer's computer, either
using his own software or that obtained from the manufacturer of the
router or a specialized software house.
Contrary to popular
superstition, simply having these reports available will not reduce
costs. Only by using bills and reports actively to make people
responsible for their calls and to manage the system will long-range
economies be obtained. If bills and reports pile up in the corner,
or are signed off without question by management, they will not do
their job. Of course, if the cost of checking is more than the cost
of abuse, you may not need the system in the first place.
Summary and some conclusions
Many modern PBXs handle
restriction quite well and, basing their operation on extension
class marks, they can restrict as a function of the calling
extension. Older PBXs, however, can often profit from use of a
stand-alone restrictor.
Passive recorders of the
stand-alone type are also good for use with older PBXs; their need
to interface with all lines (and/or trunks) tends to make their
installation relatively complex, but, once in place, they offer a
wealth of information unobtainable in any other way. Interfaced
recorders, usually reached through an RS-232C data interface, are
much easier to install; one simply plugs them in. They require,
however, that the PBX format the call data for them and present it
in suitable blocks. Such recording systems are particularly suitable
for very large PBXs where a full size computer is required to
process the call data at the end of the month. In smaller systems,
data cassettes rather than 9-track computer tapes are often
sufficient and are less expensive. The trend at the moment is toward
floppy disk systems for data recording.
Once a floppy disk
system is considered, standalone recorder-processors became
attractive. These systems sort the information and provide printouts
directly without the use of external computers.
Stand-alone toll routers
are more complex, and usually include call data recording. Sometimes
they, too, can process the information to provide reports, but
usually they simply provide data for input to another computer. All
stand-alone routers require the user to dial an identity code of
some sort; built-in PBX routing programs can access station identity
and class marks independent of user actions. Authorization codes or
billing codes can also be dialed in if desired, to permit one person
to make calls from someone else's phone.
Off-hook queuing is
easiest to implement, either as part of a PBX or a stand-alone. The
system does not have to go looking for the caller when the queue
times out or a facility comes free: the caller is right there
waiting. This minimizes holding time on the outgoing trunk.
Unfortunately, unless relatively long queue periods are used,
queuing does little to improve the occupancy of small trunk groups.
Only a longer delay in queue (5 to 10 minutes) can do that. This
requires callback queuing, which is easier provided in a PBX than a
stand-alone. Unfortunately, call-back queuing has to find the
calling party when his trunk is ready, and the trunk is held out of
service until the system has found the caller or given up the search
and gone on to the next. If the calling person is busy on another
call, the system calling back may deliver the outgoing call to a
secretary or someone else who has no idea what call is being
offered. Off-hook queuing avoids this, but a PBX, unlike a
stand-alone, can in principle be arranged not to hunt or forward
call-back calls.
With automatic routing
built in, restriction is not used as a separate feature; it becomes
part of the routing program. However, with a stand-alone, it may be
necessary to use restriction in the PBX to force calls via the
router. If the stand-alone fails, this restriction should be
removable to permit continuing communication.
A PBX with built-in
routing has an advantage over a stand-alone in that it must handle
all calls anyway, and can be used on message unit and short-haul
toll via FX lines. Usually these short-haul calls are kept out of
the stand-alone to keep its size and cost within bounds. However,
with the great increase in cost in intra-state calls brought on by
competition, the ability to handle short-haul traffic is vital.
Stand-alone units are
very sophisticated, and usually do a better job than built in
programs. However, in comparison with a modern PBX, they can very
seldom show enough saving to justify their purchase. One must use
other means to see if their better management information is worth
the extra cost and sometimes degraded transmission.
Once PBX designers
become aware of the importance of cost control and the advantages a
PBX has in providing this service as compared with stand-alone
equipment, they, too, will be able to do remarkable things. As it
is, there is almost no way to show appreciable savings of a
stand-alone against a modern PBX or a Centrex with flexible route
selection.
Older PBXs, and some of
the simpler, less expensive modern PBXs, can well afford to leave it
to the customer to add or omit stand-alone equipment. Fortunately
for the stand-alone manufacturers, this market is huge.
[
Top ] [ Next ] [
Table of Contents
] |
