Voice
Communication in Business Volume 2
Essays on telecommunications,
1981-2002
In
October, 2001, Fred Knight sent out to several of us long-time
contributors a sample RFP for an IP PBX. We were to look it over and
make suggestions, so that some real vendors could respond to it at
the upcoming Business Communications Review seminar, VoiceCon
2002, to be held in Washington, DC, Feb. 23-28, 2002. I wasn't able
to help much, mostly because my copy of the RFP came through the
Internet garbled; it seems that older versions of Microsoft Word do
not recognize documents in newer versions.
But to
finish off this section on PBXs, I built on my partial response to
Fred to show how the world has passed me by.
A Voice From The Past
(2002)
Back in 1978, I wrote
"Those Awful PBX Proposals," a humungus article which Business
Communications Review published in four parts [Ref. 1]. Then, as
now, people trying to do a professional job of choosing a PBX based
on their companies' needs, carefully described in a request for
proposal (RFP), were thwarted by vendor proposals which usually
failed to provide a clue as to what was being offered and how it
served the customer. As a one-time switching system designer, later
a representative of a large PBX customer, and finally a consultant,
writer and lecturer on such fascinating topics, I was trying to
light my little candle rather than curse the darkness.
Anybody got a
flashlight? Today, with IP-PBXs, we're in the same mess we were 25
years ago when I asked: If it costs twice as much and lasts half as
long, how can it be cheaper? The answer then was: Because it uses
"state of the art" technology [Ref. 2]. But, because few at that
time knew how to make a "lease vs. buy" decision to compare
continuing to rent from the telephone company vs. buying from an
interconnect vendor, price wasn't really that important. What turned
out to be more annoying was the way the customer found out after
cutover that his new, expensive "state of the art" PBX wouldn't
provide the minimum features and functions that had been standard
for years with relay-based systems.
Eventually this became
obvious even to some of the genius system designers, and a few
companies made some of the more necessary additions and
modifications (call detail recording, multi-button electronic sets,
easy set relocation, management access, etc.), increasing both the
already high purchase price and the cost of the maintenance
contract. Although this put a numerical value on what the customer
wanted, multi-button telephones that cost more than a color TV and
idiosyncratic administration procedures tended to leave a bad taste
and slowed the euphoric concept of "owning your own PBX" instead of
"paying all that money to the telephone company."
Today, we are facing a
somewhat similar time of decision. IP-PBXs are thrusting themselves
onto the market, people are told they should want them,*but it is
not clear what users will get at cutover. Even after attending
Business Communications Review's VoiceCon 2002, where various IP-PBX
vendors responded to a generic RFP, I still have no concept of the
architectures of IP-PBXs, nor do I know what features they have or
what functions they can perform. And I used to be an expert!
[* Footnote: But unlike the 1970s, the
Government is not forcing them to buy
—
at least not yet.]
Why IPBX?
Why an IP-PBX (or IPBX
for short) in the first place? Considering the success of Local Area
Networks, and the fact that many IPBXs are being developed by LAN
impresarios, a new "converged" Internet-compatible comm system might
more appropriately be called a Voice/Data LAN (VD-LAN). But whatever
it is called, it is what the Internuts want to sell and, like the
electronicators of 3 decades ago, allows them to insist that
customers should want "the latest state of the art technology"
whether it meets their needs or not.
The rationale for an
IPBX is based on the growth-rate of data on the Internet, allegedly
so much larger than that of voice that soon voice could be little
more than a minor increment to the load the Internet is already
carrying. For instance, one expert [Ref 3] says data today occupies
5 times as much bandwidth as voice, but not until 2010 will Internet
revenue (not profit) equal voice revenue. Presumably the VOLUME of
data will increase while voice REVENUES will continue to decline
[Ref 4] until this happy equality obtains, assuming, of course, that
at least some of these high-volume byte-bearers are still in
business. There is, after all, a difference between revenues and
profits.
Is IP Real?
The dominance of data is
an interesting thought, but is it real? We have no idea what the
COST of handling data on the Internet actually is because of the
Enron/Global-Crossing theory of economics: a company doesn't have to
make money; all management has to do is make the stock go up so they
can unload their options on the suckers. Thus we have no way of
knowing what the INCREMENTAL cost of handling voice on the Internet
of the future will be, although it is obviously ASSUMED to be zero.
Under these circumstances, with 20 to 50 times as much data bringing
in less revenue than voice brings in today, will there ever be any
profits? One wonders if moving electronic information will be worth
the investment it takes to do it.
Whatever price and
profits actually are, the vast glut of optical fiber, available at
low cost in bankruptcy sales will warp market pricing for the next
20 years, creating a false window of opportunity, even if we assume
somebody out there actually has a serious business plan. How much
will anybody be willing to spend to use more efficienly something
availabvle in vast excess?
The Flaw In The Ointment
There is, however,
another major flaw in Internut reasoning: using a TRANSMISSION
argument to justify a radically different method of SWITCHING in PBX
and CO switches. Presumably, this fallacy is based of the term
"Packet Switching" used by the Internuts to refer to the statistical
multiplexing of occasional bursts of low-speed information from
several sources to fill a high capacity TRANSMISSION line.
Just because
high-capacity Fiber Optic transmission lines are available at
ridiculously low prices does NOT imply that machines which switch
information from one transmission line to another will also be
inexpensive, particularly if such machines, when they are to access
voice, will have to perform a vast array of services which data has
never needed.
Switching systems are
different from the trunks they interconnect. While most transmission
facilities have a high basic cost per route mile, the incremental
cost to add more circuits to the basic structure is very low. With
switches, because the number of possible interconnections increases
as almost the square of the number of ports, the incremental cost of
increasing the number of ports tends to go up, not down, with size.
And, at the moment, there is no vast "unlit" quantity of digital
switches in place, waiting to be exploited, as there is with
transmission fiber.
Those offering IPBXs
insist they can build them cheaper than conventional PBXs. Perhaps
using a LAN as a switching matrix is cheaper than a time-slot
interchanger, but its much greater per-voice-line complexity (echo-canceler
+ compressor/decompressor + packetizer/depacketizer in addition to
the standard BORSCHT functions, voice/data/control muxing, etc.),
suggest that low cost voice is not necessarily on the LAN side
unless most of this cost can be moved to the user terminal (IPfone?)
which the customer has to buy separately.
But even assuming
hardware costs for an IP-based PBX system are less than those for
conventional switches, about 70% of the actual cost of developing a
PBX is in the software. Legacy PBXs already have most of that cost
covered and written off; the data guys have to start from scratch.
Maybe they can write such programs by osmosis, and maybe they can
intuit all the things PBX programmers have spent 30 years learning
the hard way, but considering how badly most PC software works,
particularly for communications, I doubt it.
Software Bugs
One would suppose the
Lucents, Nortels, Fujitsus and others who already have highly
developed legacy PBXs would be in a better position to add LAN
features than the LANcers to add PBX features. The former, at least,
had a long record of profitable operation prior to being bogged down
in "the new economy," while the latter had little experience with
profits on their own turf before they thought of adding complex PBX
requirements.
Even so, when both the
legacy and new-age designers THINK they know what features and
functions are important to system users, we can't trust their
optimism until some real user experience with these new products is
available. The whole exercise is meaningless if the product can't do
what the customer needs, or if the vendor cannot generate enough
cash-flow to stay in business to provide after-sale support [Ref.
5]. Although Judge Greene ruled that potential PBX customers could
not inquire into the financial stability of interconnect PBX
suppliers [Ref. 6], that was a long time ago. Recent experience
suggests that anybody who doesn't investigate a vendor's financial
stability deserves whatever he fails to get.
Phone Calls In The Phuture
To explore further any
effort to justify "Why IPBX?" we should keep in mind that for the
foreseeable future, the great majority of telephone calls will
continue to be entirely local: either internal to the PBX itself, or
to nearby customers of the serving central office and other COs in
the local calling area. Replacing almost new DMS-100 and 5ESS CO
switches with some form of as-yet unproven IP voice switching
systems (which will have primarily 2500 type telephones as
terminals) will not take place overnight. Thus the theoretical cost
advantages of IP for long distance transmission and the real
disadvantages of IP as a voice-switching fabric for local calls
suggests the best buy might be a gateway for one or more DS-1
channels from a voice PBX to the Internet, perhaps as a part of the
Router for the associated LAN.
Even then, one wonders
if the Internet can carry LD voice at a profit when business
customers today can get a better deal than the 3 to 5 cents a minute
offered to residential customers.
Voice And Data: Two
Differences
As for regular PBX
features under IP, we have two philosophic difficulties. First, the
telephone system generally makes point-to-point real-time two-way
connections, while the Internet generally makes a series of one-way
meet-me connections. That is, when I phone you or you phone me, a
two-way connection is established and we interact directly with one
another. Using the Internet, however, I usually send a message to
your e-mailbox where you eventually retrieve it and respond by
sending a message to my e-mailbox, or else we meet in a chatroom and
leave messages for each other, almost in real time, on the
chat-rooms server. Perhaps more often, we call up a data-base and
download files that someone else has stored at some time in the
past.
Presumably IPBXs will be
able to set up two-way point-to-point and multi-way real-time voice
and video connections on demand, in addition to simple data
transfers. But a two-way real-time information flow is not what the
Internet presently provides.
In the second
philosophic difference, the Internet concept is supposedly based on
eliminating any form of central control while "migrating
intelligence to the edges" of the network. By contrast, voice
switching traditionally requires a central control for each switch
to allocate resources and establish connections. Further, customer
features on PBXs and CO Switches require extensive data bases to
handle user-specific features such as call waiting and distinctive
ringing, group features such as hunting, pickup, screening,
conferencing, message center, etc., and over-all features such as
call detail recording (CDR) and automatic route selection (ARS).
Obviously, the Internet
has a certain amount of central control to manage translations
between user name, permanent identity of the network interface card,
and currently assigned Internet address. Similarly, PBXs and CO
switches often have distributed control such as the customer's
personal calling list stored in the phone, and various displays and
alerting functions under direct user control. Further, distributed
control of switching reached a peak with Almon Strowger's Step by
Step system, where each digit the customer dialed operated a
different switch to complete the call.
Never-the-less, the
whole control philosophy of associating one LAN terminal with
another, and finding paths through the Internet itself is quite
different from that used by PBX and CO circuit switching. Although
both use the concept of a "destination code" as a more or less
permanent identification of a user and a "route code" which is
applied at switching nodes to find a specific outgoing path for an
incoming signal seeking a partiacular destination, there are obvious
differences. Packet switching grew out of the message switching of
teletypewriter networks, and works best where the entire message is
one packet with one header identifying the called terminal. It will
be at something of a disadvantage when it is used essentially for
each syllable uttered in each direction, as with VoIP.
However Internet address
translations and routing evolve, one would suppose an IPBX would
have its own central data base for things like busy-idle terminal
status, hunt and pick-up groups, etc., as well as smaller data bases
in individual PCs for personal dialing lists and the like. Each IPBX
should be able to handle any kind of call set-up, incoming, outgoing
or intra, including response to busy and no-answer, etc. The only
problems with distributing too much control to individual PCs,
besides the endless duplication implied, is keeping scattered data
bases updated, and obtaining billing and traffic data from a
multitude of repositories. Perhaps IPBX designers will pretend the
box in the back room is a Voice-Data Router, put the local system
data base there, and continue to pretend there is no common control.
What Do I Know?
So let us suppose that
something called an IPBX can be built at a price competitive with
existing PBXs, and IP designers are just as smart as the electronic
PBX designers of 30 years ago who had also never been in a business
office and who also knew absolutely nothing about how businesses use
the phone. Am I smart enough to offer them or their potential
customers definitive advice? Although at one time I knew all the
tricks of doing station surveys, how to read equipment inventories,
and what questions to ask, particularly of the switchboard
attendants and the senior secretaries who really knew how their
companies worked, I myself haven't been in a business office in ten
years. What would I have to look for today, as opposed to the 1970s
and 1980s? What are the changes that have taken place behind my
back, making me obsolete?
The first thing I would
do if a company wanted me to prepare an RFP for an IPBX for one of
their locations would be what I have always done: review their
existing phone and data costs including such esoteria as floor-space
requirements and any special switch-room air conditioning needs,
particularly in winter. I would want to get an idea of how much is
being spent today so that I would have an upper bound on what such
costs could be with the new system in place.
Evidently, if LD costs
are already 3 cents a minute or less, anywhere in the US, I would
not need to specify elaborate Automatic Route Selection (The two
exception might concern intra-LATA and intra-state calls, and calls
outside the US; in both instances, prices might be irregular and
much higher). I could also question the depth of detail needed for
Call Detail Recording, and the type of restriction patterns, if any,
required. However, some companies such as law firms use the
telephone detail for billing of professional services, while
telemarketing centers need to know where their incoming calls are
coming from and all the various parameters required for ACD
management.
One particular calling
pattern I would investigate with great care would deal with Fax
service. Which are the Fax lines and how much of the calling cost is
attributed to these lines? What are the destinations of Fax calls?
Can these destinations be reached via the Internet, or will it be
necessary to continue to use regular telephone connections to reach
vendors, customers and peers directly rather than through an
intermediate storage system? I would also be interested in costs for
access by telecommuters, both voice and data.
Next, I would walk
through the client's offices and see what equipment is actually in
place today and how it is used. I strongly suspect that the three
biggest changes I would see from ten or twelve years ago would be
the presence of LAN-connected PCs, the actual extent of Fax use, and
the incredible penetration of cordless phones, both PBX and
cell-based. What I would not see, but would have to ask about, would
be the presence and use of voice mail (VM).
The latter, of course,
impacts the way telephone groups are organized. Not too long ago,
all calls to executives, and to many higher level working people,
were screened or at least picked up on ring-no-answer by the
department secretaries; a call to the "front door" number was
screened by the console attendant as well. Today, most calls seem to
go DID to the recipient directly, and are forwarded to VM in case of
busy, no answer or don't interrupt; only then do voice prompts offer
the caller a chance to leave voice-mail or, perhaps, reach an
alternative human.
Another approach has the
caller reach an Automated Attendant (AA) directly where an
impossible voice menu is offered, usually including the privilege of
keying in via Touch-Tone the called extension. This preserves the
savings of DID but allows the company to have a single directory
number and regular trunks. I would, first of all, have to know how
the AA, the VM, the secretaries and console attendants (if any)
divvy up incoming calls, and how secretaries, in particular,
participate in setting up outgoing calls. With present calling
patterns in mind, I could then ask what the management wants to do
with a new system: same, or a new and improved approach?
The next thing I would
look at is who has computers and how they are used. Does EVERYbody
REALLY have the very latest Pentium with the very latest
Microsoftware? Or are there Luddites who haven't yet learned the
mysteries of this week's version of Windows, graphic designers who
insist on Macs, and some people who, for one reason or another, have
no computer at all? Is it company policy to update every time a
computer or software vendor needs to improve his cash flow? Is
"sneaker net" still being used in spite of or because of the
presence of a LAN? Exactly how does the LAN presently connect to the
Internet? Will the new system be expected to combine e-mail with
voice mail? Will the system have to read e-mail to a telephone, or
transcribe voice-mail into computer text?
Computers And Telephones
Obviously, for those who
have a computer, an IPBX might expect to use the capabilities of the
computer's display to minimize the complexity of the telephone set.
The ideal might be a sound-card with a jack into which a standard
telephone hand-set (or headset) could be plugged, the rest of the
telephone's innerds being on the sound-card. The handset (or
headset) serving as microphone and private speaker would be helpful
in Dilbert-type cubicles.
The required display and
control functions would be in a software package that could take
full advantage of the computer's monitor, keyboard and mouse,
perhaps emulating a 1A2 phone with a pop-up picture showing line
pick-up buttons and lamping, along with a callING and callED number
and name display, automatic dialer, and voice-mail control. Indeed,
a similar sound-card/software package could be used by conventional
PBXs as well as those for IPBXs, assuming suitable communication
between the PC and the PBX and VM controls could actually be made to
work.
Using the PC in this way
brings up a very important point: what happens to telephone calls
when a user's computer is turned off? Perhaps a separate telephone
set, program-related to the PC should be used, rather than a just a
handset that is part of the PC. Then, as long as the IPBX knows when
the PC is turned off, it could treat the phone as a separate,
simpler entity. But then, should the phone be a simple 2500 set with
a special line-card, or an IPfone which can interface the IPBX the
same way a PC interfaces a LAN? Obviously, several levels of IPfone
could be provided, particularly for those who are heavy phone users
but who do not need a PC, but short of having a standard IPfone
available from a variety of vendors, this would add a great deal of
cost and time to the development of an IPBX, to say nothing of cost
to the user and the cost of user training in perpetuity.
The General Power Problem
The power on/off problem
has other aspects. Traditionally, PBXs (and CO switches) provide
power to their own phones, usually via the same pair of wires that
carries speech and signaling. This seems a little impractical if the
terminal is a PC.
When PBX phones are
compatible with CO trunks, power-failure (or system-failure)
transfer can be used to connect selected telephone sets directly to
CO trunks. This is particularly useful when large back-up batteries
or diesel-electric power systems for emergency use are impractical.*
[* Footnote: Charging batteries give
off hydrogen which is highly explosive, and a diesel-powered
generator on the 50th floor of a skyscraper seems to offer a variety
of problems including noise and pollution. Perhaps fuel cells will
be the answer.]
Although electronic
multi-button sets are often powered from their PBX, they are usually
incompatible with conventional CO trunks and, when DS1 or PRI
channels to a CO are used, even digital telephone sets are not
compatible. It seems likely, too, that IPfones would not be
compatible with either analog or digital CO trunks. Thus separate
phones terminating directly on the CO might be provided for
emergency use. Fax lines, if used primarily for outside connections,
might well serve for voice calls in emergencies, assuming an
associated telephone could operate in the absence of AC power.
When emergency power is
provided (as in hospitals and other institutions), the amount
available for communications may be limited. It is easy enough to
keep a PBX which powers its own 2500 sets going, but I would check
carefully to see if there is enough emergency power available for
PCs and all their peripherals. Perhaps it would be better to give
each a small UPS which could hold over momentary power glitches, and
allow computers to be shut down in an orderly manner in the face of
a longer outage.
This, at least, is the
sort of thing I would look for. But I would have to bring in an
associate thoroughly familiar with real-world Internet operations to
make sure whatever special features or functions which might come
from the special properties of the Internet would not be missed.
Conclusions
Although it is certainly
possible to make an IPBX, it is also likely that, if such a machine
includes the cost of its terminals, it will NOT be any less
expensive than a legacy PBX and its terminals. Further, the addition
of necessary PBX features to something intended for data switching
will have a large programming cost and will require a more
flexibility than appears to an outsider to be necessary for a LAN.
The one place where an
IPBX might be able to save money is in long distance VoIP, voice
over Internet protocol. This is described as particularly useful in
overseas calls where government-run telephone systems have
artificially high prices, but seems to be less attractive compared
to the rates a business can negotiate with conventional IXCs. It
would appear that local calls cannot take any advantage of VoIP
until LECs are serving terminals that are mostly IPfones.
Further, until actual
costs, independent of fire-sales and subsidies, are known, and some
stability returns to the telecom market so that after-sale support
can be considered dependable, the use of IPBXs will be more of an
daring adventure than a practical business arrangement.
REFERENCES:
1. Goeller:
Those Awful PBX Proposals. Business Communications Review,
MA, MJ, JA, SO 78.
2. Goeller: The
Art of the State. TeleConnect, Sept. 84.
3. Lawrence
Roberts, quoted in Barthold: Broadband the Only Salvation for ISPs
Nearing Extinction. Telephony, May 27, 2002.
4. Krapf: Voice
Services Pricing: How Low CAN They Go? Business Communications
Review, July 02.
5. Goeller: The
PBX and Downstream Dollars. Business Communications Review,
Sept. 89.
6.
Electronic News, July 2, 1984. p27.
[
Top ] [ Next ] [
Table of Contents
] |
