Xem mẫu
- Networks and Telecommunications: Design and Operation, Second Edition.
Martin P. Clark
Copyright © 1991, 1997 John Wiley & Sons Ltd
ISBNs: 0-471-97346-7 (Hardback); 0-470-84158-3 (Electronic)
45
Corporate Networks
Although large multi-national companies can afford to employ specialist ‘telecommunications
managers’,smallercompanies individuals
and are less privileged. For theselatter
groups,
telecommunications are often just another subsidiary responsibility either a computer services
for
manager or even the general business manager. Nonetheless all telecommunications managers can
affectthewayinwhichtheircompaniesoperate.Inthischapterweshalldiscusshow.The
following topics are covered
0 telecommunicationsperformanceandmanagement
0 cabling for telephonesandcomputers
0 officecomputernetworking(LANs,etc.)
0 privatenetworkdesign and operation
45.1 TELECOMMUNICATIONS MANAGEMENT
The telecommunications manager should manage all aspects of telecommunications.
He needs totake ownership of the full cost andfunctionalperformance of all
information flows, by whatever means, throughout his company. The job should not
consist only of designing and operating the company’s private networks. This loses
sight ofthe‘strategichorizon’,theprofound effect that effective all-round commu-
nication can have on a company’s operations.
Telecommunications management involves the day-to-day operation and adminis-
tration of resources, and it should include creative and strategic questioning.
0 What is our long term business goal?
0 What are the most important ‘information flows’ supporting it?
0 Why is ‘key’ information being carried on a particular medium?
0 What is the capability of new technology and how can it affect the core business
processs to advantage?
815
- 816 CORPORATE
0 Why should I maintain a private network rather than use public services? Which
gives better ‘added value’?
0 Why are we spending so much money on paper mail?
0 Why are we using facsimile rather than electronic mail?
0 How can we gain competitive advantage?
0 How can we lock-in our customers using telecommunications?
0 Which are the right standards for me to use?
0 How will legal regulation affect me?
0 Above all, are our board and seniormanagementproperlyclued up about tele-
communications? If not, is that not my fault?
Good telecommunications management keeps a company using the right technologies,
the right suppliers, the right control procedures and the right monitoring measures,
maintaining optimum performance at all points in time.
From the suppliersthere is agreaterneedforproductswhichencompasswhole
business solutions, tailored packages of services, products and support from a single
supplier to meet the need of a business process neatly completely. No longer has the
and
telecommunications manager got the timeto bind together a rag-bagof bits, essentially
the solution himself. Neither can he afford the time needed to resolve any subsequent
problems or take on any necessary enhancements. What he has to do is to identify
the need, understandit completely and find suitable suppliers.
Thus he seeks a
communications solution for ‘order collection’, ‘customer payment’, ‘customer account
management’, etc.
Having found solutions to the business communication needs, a framework neededis
forperformanceand overallcostmanagement.Wecoveredthisinsomedetail in
Chapters 34 and 35. Finally, the manager may turnto the technology and to thedesign
andoperation of networks.Theremainder of thechapter, discussessome of the
important current issues facing telecommunications managers in this particular area of
responsibility
0 building
cabling
0 computer
networking
0 privatenetworkdesignandoperation
45.2 PREMISES CABLING SCHEMES
Until the explosion in the number of office computer workstations in the early 1980s,
buildings were designed and pre-wired with the needs only of electricity and telephone
cabling in mind. However, the large number of personal computers, word processors,
facsimile
machines,mainframe computerterminals,
printers and electronicmail
terminals now in use means that the needs of data terminalcablingnowgenerally
exceeds those of simple electricity and telephone.
- PREMISES CABLING SCHEMES 817
At first, companies installed
dedicated wires for each terminal as they were
needed; installation work and the costs associated with it mounted on a‘pay-as-you-
up
go’ basis. In time the wiring became a complicated spaghetti of leads strewn across false
ceilings, knotted on cross-connection frames, and squeezed into congested conduits.
Eachtimea new cable was needed,theceiling tiles wouldcomedownagain,the
conduits would be opened up, and somehow another path would be found. The costs
began to grow out of proportion.Each new cablewasharder to install thanthe
previous one and each new installation led to faults being inflicted on existing wiring as
cables were drawn into already congested ducts.
Most companies now favour formally structured cabling schemes which are planned
and installed at thetime of initialbuildingdevelopment orduringrefurbishment.
Sufficient conduits, equipment cabinets and cables are provided to cater for wiring in
excess of all foreseen needs.
Figure 45.1 illustratestypical
a structured cabling scheme. It comprises cables
running out from a main trunking room or patch frame. First a number high-grade of
backbone or riser cables run out (inbuildingconduitsoftencalled risers) to patch
panels (passive components) or to cablinghubs (these are ‘active’ components which
may convert electrical formats or interfaces as may be needed) located on a one-per-
floor basis. From these secondary locations, a large number of lower grade cables run
out to all possible locations withinthe office floor. It is common nowadays to find glass
fibre cables used in the risers and unshielded twistedpair ( U T P ) or shielded twistedpair
( S T P ) copper cables used within a given building storey.
Typically four four-pair category 5 (Cat 5 ) cables (i.e. eight wires per cable) are run
from the floor patch panel to each single occupant office and terminated on eight-pin
RJ-45 sockets, and four RJ-45 sockets per office allow for up to two telephone exten-
sions and two data terminals per desk. Standard telephone guage unshielded twisted
pair ( U T P ) is suitable for most LANs, data and telephone needs. For higher perform-
ance needs, shielded twistedpair ( S T P ) ,coaxial cable or fibre optic cable may needed
be
(see Chapter 8).
Thefloorpatchpanelprovidesalltheconnections onthat floor andforlater
reconfiguration simply by changing the patch wire. Local area networks ( L A N s ) of
logical star, bus and ring topologies can be created and reconfigured at will. In addition
the floor patch panel is connected via riser cables to patch panels on other floors and
back to the main patch frame. These allows for connection to public telecommunica-
tions networks and the mainframe computer installations on other floors. The riser
cables are usually either shielded twistedpair or fibre.
Though structured cabling schemes require greater capital outlayfirst, they usually
at
reward their owners with much lower running costs, very fast and cheap terminal wir-
ingleadtimes and muchreducedfrequency of faults.Thelargest single cause of
wiring errors, tampering with the cables in the conduits, is eliminated by a structured
scheme. Unstructured (ad hoc) wiring schemes should thus only be considered where
very low terminal populations exists (less than about 20 terminals).
Structuredwiringschemesgenerallyrequire very littlemaintenanceandoperate
virtually trouble-free, but good management practice demands that a proper inventory
be kept of the use of each wire, and that each be properly labelled.
A number of regulatory stipulations need to be considered during the planning and
installation phases cabling. First, it not good practice to run cables in air conditioning
of is
- 818 CORPORATE
Floor
patch individual
offices
to
Wires
panel
Plentifulsupplyofunshielded
twisted cable
pair
Computer workstations
- Floor
patch
in telephones and
office
locations
panel
Wires
rl- Floor
potch
Wiring cabinet (one
multiplexer other
and
per floor)
equlpment
t o individual offices
- may hold
Shieldedtwisted
pair
C orfibre
r Main
patch
frame Company mainframe computer
services U
Figure 45.1 Structured cabling scheme
ducts because of the smoke and fire risk arising from the combustion of the plastic
insulation. (Actually, United States permits this but it illegal in the UK). Separate
law is
conduits for electrical and telecommunicationswiring are recommended to protect tele-
communications users and equipment alike from the dangers of a mains electrical ‘shock’.
- RKING OFFICE COMPUTER 819
In Germany,fire regulations require sealing up the inside of the riser conduits at each
floor level with cement. This makes the pre-installation of many cables in a structured
cabling scheme attractive, as the effort and cost associated with drilling through each
floor, pulling a single cable and then re-sealing the cement would be very expensive and
time-consuming.
However, a regulatory restriction applies in the United Kingdom which may cause
companies to considerseparatecabling schemes fordataand telephone uses. The
restriction is that all cables and patch frames carrying telephone wiring which may
possibly be connected to thepublic telephone network(via the PBX) must maintained
be
by the registered PBX maintainer. Furthermore, any data or other equipment connected
to wires in the same cable (irrespective of whether these wires are connected to thePBX)
must either be approved for connection to the telephone network (with ‘green dot’),
or they must be isolated by means of a barrier box (intended to prevent high voltages
leaking onto the public telephone network). These restrictions are easy to avoid by
installing separate structured cabling schemes for telephone and datawiring. Thus cables
and patching equipment are duplicated, but conduits and cabinets may shared. be
45.3 OFFICE COMPUTER NETWORKING
Issues in the design and management of office computer networks are
0 meeting capacity and response time needs
0 flexibility for reconfiguration
0 compatibility of application and network management software
We discussed network dimensioning to meet data network capacity and user response
needs in Chapter 30. Typical response times needed from a real computer network are
shown in Table 45.1. Achieving such targets takes a considerable amount of skill; there
are far too manydetailed factors to be considered which we have not the time for here,
but many excellent books,consultancyorganizations and suppliers are available to
Table 45.1 Typicalresponsetimeexpectations
‘Maximum’
Function response time
Computer activation 3.0 S
Feedback of error 2-4 S
Respond to identify code 2.0 S
Keyboard entry 0.1 S
Respond to simple enquiry 2.0 S
Request for next page 0.5 S
Respond to ‘execute problem’ 15.0 S
- 820 CORPORATE NETWORKS
assist with complex data networking problems. Basic guidance is easier: ‘keep it as
simple as possible’. Careful design of relevant software, together with the use of the
formulae given in Chapter 30 will go a long way!
The flexibility to reconfigure networks to
data or meet a number of needs
simultaneously is derived from the use of a number of tools and techniques, e.g.
-
e structured
cabling
e localarea networks ( L A N s ) (Chapter 19)
e open computer architectures (e.g. OS1 and TCPIIP, see Chapters 9 and 19).
The use of L A N s or other computer network architectures (e.g. X.25, framerelay,
ATM, TCP/IP, APPN, SNA), needs to be tempered by careful preparation. None of
the tools are as flexible as they may seem, and lack of appreciation of this fact can lead
to equipment incompatibilities.
Local area networks ( L A N s ) should be designed and run using rigorous ‘mini-data
centre’ procedures. Out of hand, they canbe very difficult to manage and problems can
be hard to diagnose.
Theselection of aparticular LAN (e.g. ethernet or tokenring) oraparticular
computer architecture (e.g. TCPIIP, OSI, S N A or A P P N ) needs to take cognizance of
the applications to which itwill be put (e.g.$le transfer or printer sharing or mainframe
database repository) and also of the
computerhardware software
and (e.g. PC
operating system software:DOS, Windows,Windows95,WindowsNT, OS/2, UNI,
M V S , etc.)whichit is expected toserve.Equipmentfromaparticularcomputer
manufacturer tendsto have been developed one
with architecture mind.
in For
example, the IBM architecture is SNA or A P P N and the preferred LAN technology is
token ring. Meanwhile Digital Electric Comporation’s( D E C ) proprietary architecture is
called D E C N E T and the preferred LAN technology is ethernet. In the longer term,
internetprotocols (e.g. T C P j I P ) and open systeminterconnection ( O S I ) standards
allow greater inter-changeability of computer equipment and network components.
45.4 PRIVATE NETWORKS
Public telecommunications networks are usually large and complex, requiring massive
capital and manpower resources. a result, network changes are usually conducted in
As
a careful and steady manner because the cost of mistakesis magnified by the vast scale.
Also, because public telecommunications operators (PTOs) are normally required by
terms of licence to provide services of uniform availability over widegeographic areas,
the development and rollout of innovative and technologically advanced services can be
held up, and prices to main business customers can prejudiced by the subsidisation of
be
uneconomic rural areas. Because of this, and driven by the desire to adopt the latest
techniques, some companies have developed extensive private or corporate networks.
The advantage to be gained depends on the service offerings and tariffs of the public
telecommunications operator(s) and also on the legal constraints placed locally on such
a network.
- PRIVATE NETWORKS 821
Site 2
Customerpremises
equipment ( C P E )
Site 3
Figure 45.2 A privatetelephonenetwork
A typical private network is shown in Figure 45.2. The example shows threeprivate
branch exchanges (PBXs) in different office buildings, interconnected by circuits
between the buildings. The PBXs and other customer premises equipment ( C P E )may be
owned by the private company, or leased from the PTO. The wiring in each office
normally belongs to the private company. However, the circuits interconnecting the
different PBXs in different buildings almost certainly are leased from the PTO. Thus a
private network usually comprises customer premises equipment (or customer apparatus)
and leased circuits which interconnect the different locations.
The private network illustrated in Figure 45.2 is capable of switching telephone calls
between any two telephones in any of the three different offices.If desired, a single
numbering plan could cover all three offices, each telephone having a unique three- or
four-digit extension number. In this way users benefit from a short dialling procedure,
and the company can savemoney by transportinginter-officecallsoverthe leased
circuits rather than having to pay the full public telephone tariff for each call. Provided
that the telephone trajic between different offices exceeds a given threshold value (the
exact value of which depends on the relative leased circuit and PSTN call tariffs) it will
almostcertainly be cheaper to use direct leased circuits this
in manner, asthe
calculation in Table 45.2 shows. The cost benefit of the private network (compared with
using the public network) is prone to changes in PTO tariffs. As an example, in the
United Kingdom the balance of public service tariffs and leased line rental charges has
changed so radically since the early 1980s that many companies
large are now
considering the abolition of the private networks they established at that time. Another
factor encouraging companies to give up their private networks is the desire to reduce
the in-house manpower needed for the day-to-day responsibilitiesnetwork
of
operation. So the pressure is for the public network operators to assume this greater
responsibility!
- 822 CORPORATE NETWORKS
Table 45.2 Threshold for leased circuit consideration
Leased circuit tariff = EL per quarter
Public telephone tariff per minute = & p
Telephone usage minutes per quarter = m
Public telephone bill per quarter = mp
therefore if mp > L
or minutes per quarter m > L / p
a leased circuit is worth consideration
The calculations shown Table 45.2 are oversimplified. The full analysis should take
in
account of traffic profile effects (see Chapters 30 and 31). Only the point-to-point traffic
should be considered, and the numberof leased circuits costed needs to take account of
the grade of service (following the Erlang formula).
Private networks are common in medium and large companies where the economics
of large scale helps them to ‘pay-in’, particularly for intra-company communication
between main offices. For communication with suppliers or customers, or even to staff
in small remote offices, private networks are often connected to the public switched
telephone network (PSTN). In Figure 45.2 each of the PBXs is illustrated with a direct
link to the PSTN.
Apart from cost or the capabilities of new technology, another motive for a company
to establish a private network could be the network security afforded. Take a small pri-
vate packet-switched or other data network which connects a computer holding sensitive
data to a number of remote workstations. Unauthorized users with workstations which
are notconnected to the private network cannot gain access to thesensitive information.
When it is not possible or economic to establish a private network to ensure the
security of information, scramblers or encryption devices may be employed in conjunc-
tionwith the public network. Scramblers work pairs,
in at eachend of acon-
nection, converting speech into a coded, but unintelligible signal for transmission on
the public line. Equivalent devices when used to scramble sensitive data are usually
called encryption devices.
45.5 ARCHITECTURE OF PRIVATE NETWORKS
Public and private networks are based on similar technical principles, but differ in scale
and complexity. Privatenetworks generally cater for much lighter traffic, but more
specialized service needs. An example of the difference in technical standards is
illustrated by making a comparison between the signalling systems used for conveying
telephone calls between PBXs over a private network with those used between the
exchanges of a public network. Inter-PBX signalling systems often only carry short
digit strings (equivalent to a four-digit extension number), but in addition they need to
carry information necessary for invoking special features such as ring back when free,
etc. As a result, the speed at which individual digits of the dialled number are sent
- ARCHITECTURE OF PRIVATE
NETWORKS 823
between the PBXs may not be the critical factor, because the small number of digits
constrainthecallsetuptimeanyway,butthesignalvocabularymust be wide to
support all the various special functions. By contrast, inter-exchange signalling systems
need to carry longer digit strings.A full international number and its prefix could be up
to 18 digits long, and must pass quickly between exchanges so that the set-up delay is
reasonable.Exchangesinapublicnetworkmay need to transferinformationon
charging or routing the call, as well as any customer-requested supplementary services
(e.g. calling line identity (CLZ) or ring back when free). Finally, the inter- exchange
signallingsystemmay be used to conveyinformationfornetworkadministration,
including network management information control
and signals, or
for remote
operation and maintenance commands.
Public network standards tend to be robust, designed to withstand and control the
severe and diverse demands of the wide range of uses to which public networks are
exposed. They need to be resistant both to failure and fraud. Private network standards
are nearly always derivatives of public network standards, and are often developed
from the interface used to connect customer premises equipment to public network
exchanges. Figure 45.3 illustrates an example pertinent in the UK, where the primary
rate ISDN PBX-to-network interface, called D A S S 2 (digital access signalling system
No. 2 ) is similar to the digitalprivate network signalling system ( D P N S S ) which may be
used on private networks between PBXs to support ISDN-like services (the ITU-T’s
signalling systems for the similar purposes are called DSS-l (digital signalling system I )
and Q-SIC. These are defined in the Q.33X series of recommendations). The advantage
of using similar standards for PBX-to-exchange and PBX-to-PBX interfaces is that it
minimizes the complexityof PBXs which are required to operate in both modes. Similar
benefits can be gained in other typesof networks (e.g. packet networks, etc.) aligning
by
the technical standards used in private and public network variants. Minor differences
are inevitable, have always existed and are bound to persist. They are the results of the
contrasting network circumstances and demands on public and private networks. Thus
a private packet-switched networkis likely to be based on ITU-T’s X.25 standard, but it
may include a number of customized software features. Likewise the message handling
system, as we saw in Chapter 23, recognizes separate public and private management
domains.
c) Public
Private
network ISDN
Figure 45.3 PBX-to exchange and PBX-to-PBX ISDN signalling
- 824 CORPORATE NETWORKS
45.6 PLANNING PRIVATE NETWORKS
The smaller traffic scale of private networks, coupled with their relative freedom from
licence constraints,meansthat their structurecan differfrom that of publicones.
Indeed, the optimum topology is not only independent of the real cost of equipment,
but can be highly distorted by the PTO’s relative tariffs for leased circuits and public
network services.
Besides the economy measures of Chapter 38, three particular network routing tech-
niques which offer considerable scope for cost reduction to private network planners
are
0 establishmentoftelecommunicationshubs
0 publicnetworkoverflow
0 publicnetworkbypass
45.6.1 Telecommunications
Hubs
The small traffic scale of private networks tends to encourage the use of a star net-
work topology in which one, or a small number exchanges at the hub of the network
of
provide a transit switching point between all other exchanges. The configuration is
illustrated in Figure45.4, where all calls between any pair exchanges are switched via
of
the hub exchange A. By employingsuchatopologyahandful of circuitsmaybe
sufficient tocarry alltraffic (let us say 3 circuitsfromeachoutlyingexchange to
exchange A, a total of 21 circuits), as against a more meshed or fully interconnected
networkofonly 1 circuitbetweeneachpair of exchanges which wouldrequire 29
circuits (as shown in the inset of Figure 45.4).
PBX or other
exchange
29 circuits
Figure 45.4 ‘Hub’ or ‘star network’ topology for private networks
- PLANNING 825
Thestar topologyshown in Figure45.4 is similar tothe hierarchicalnetwork
structure described for larger scale (public) networks in Chapter 32. The smaller traffic
on privatenetworks means that the use of the star topology (or hierarchical structure) is
nearlyalwaysmorecost-effective thana moremeshednetwork of inter-exchange
connections. It not only minimizes the total number circuits required, but also tends
of
to minimize the total number of circuit miles. Minimizing the number of circuit miles is
most important, because it is this value that relates to the leased circuit charges levied by
the public telecommunications operator (PTO). Careful siting of the hub exchange is
also important. In international private networks, the choice of the hub site may be
especially important, because the price of leased circuits may not be the same in each of
the countries. Let us return to Figure 45.4. It might be the case that exchange A is the
geographical hub of the eight sites, that the prices of leased circuits into and out of
the country in which exchangeF is situated are cheaper. In such an instance, it may be
more economic to use exchange F as the network hub. Similar anomalies of leased
circuit prices can arise even within a single country, caused either by a banded (rather
than directly distance proportioned) price structure or because of the differential price
structure of competing PTOs.
Circuit numbers are calculated in the normal way for the type of network and the
circuit lengths are measured as radial distances. The costs can then be worked out
accordingly.Repeating theprocedurefordifferentoptionalnetworksbasedonthe
various available hub sites helps to determine the cheapest configuration.
45.6.2 Public Network
Overflow
The traffic between any two exchanges of a network has a ‘peaked’ profile. However,
although public networks are mandated to carry this peak of demand, there is no
similar mandate for private networks. In particular, there is no need for the private
network to be able to carry the peak demand using its own resources alone. Instead it is
permissible, and may be far more economic, to overflow the peak demand to the public
network. Figure 45.5 illustrates this technique. Two PBXs (or other type of private
exchange) called A and B, and located at different offices of the same company, are
connected together as a private network. For most of the day the traffic demand between
exchanges A and B is between 6 and 8 Erlangs, but between three and four in the
afternoon it peaks at 12 Erlangs.
The private network planner has done some cost calculations. He has established a
network of eight directleased circuitson the private network between exchangesA and B,
but has elected to overflow the excess 7 Erlangs of peak traffic via the public network.
The total cost of the configuration is difficult to calculate, because the cost of calls
overflowed via the public network mustbe measured or estimated from the complicated
teletraffic formulae presentedin Chapters 30 and 32. For each hour of the day thetraffic
that is carried on the eight circuits direct route, and the traffic overflowing from it, is
determined using the Erlang formula, by inputting the traffic demand for that hour.
Thus, until 5.00 a.m. there is no traffic and no overflow. From 9.00 a.m. to 3.00 p.m. the
traffic is 7 Erlangs, and the overflow from eight circuitsis 1.25 Erlangs so 75 call minutes
per hour are sent via the public network. Similarly, during the hour of peak activity
(3.00p.m. to 4.00p.m.) the offered traffic is 12 Erlangs, the overflow from the eight
- 826 CORPORATE
Public
network
/----
l
I
Overflow
Exchange I
I A I
*
m B
I
I , B leased c i r c u y I
I I
Traffic
demand 12. Traffic overflowed
via public network
A to8
10.
8-
6.
carried over
4. 'private' network
2.
0
12 6 9 12 3 6 12
Figure 45.5 Public network overflow
circuits is 5.07 Erlangs, and 304 minutes route via the public network during this period.
Over the whole day (adding also the overflowed minutes in the 4.00p.m. to 6.00p.m.)
a total of 904 minutes overflow dailyvia the public network. At a cost &P minute
of per
and &Lper quarter tariff for each leased circuit, this configuration an approximate
gives
overall quarterly charge of
Estimated quarterly charge = ( L + 3 X 22 days* X 904 X P)
(There are approximately 22 business days per month.) As in the last example, the
optimum number of leased circuits (for minimum cost) is determined in a trial-and-
error fashion by calculating two costs for various configurations.
45.6.3 Public Network Bypass
Wherepermitted by law,publicnetworkbypassmay be anadditionalmethod of
A
deriving additional telecommunications savings. cost saving is made on long distance
calls over the public network routing thecall as far as
by possible within the private net-
work before handing it over to the public network for completion. By so doing, the
public network is bypassed as far aspossible, the shorter geographical length the pub-
of
lic network connection reducing the chargeable tariff. Figure 45.6 illustrates the method
- KS PLANNING PRIVATE 827
Final destination
Private
network
London
(company A )
..... Alternative
network
public
route
Figure 45.6 Public network bypass
ofbypass. Intheexampleshown,a company privatenetwork is establishedinthe
United Kingdom, linking exchanges in LondonandEdinburgh.A
private caller
connected to the private exchange in London wishes to make a call to an office in
Edinburgh.Unfortunatelythedesireddestination is not connecteddirectly tothe
desired network. However, routing the call over the private network from London
by to
Edinburgh, and then into the public network for the final short hop, the public network
(and itsdistancedependenttariff)has been bypassedforthegreater part of the
connection. Only the tariff of a local call is payable, rather than the much higher tariff
that would have been due if the call had been routed over the public network a trunk
as
call for the entire distance between London and Edinburgh. Whether or not the overall
economics favour such a bypass may change from day-to-day, depending on public
network tariffs and on the existing topology of the private network. You should not
assume that the example used is economic in all cases; specific evaluation is necessary
on each occasion. Furthermore, the transmission quality of the end-to-end connection
may preclude certain configurations. Particularly in thedata networking case, the speed
of data propogation is often critical.
Anothermeans of achievingpublicnetwork bypass isby the use of out-ofarea
exchange lines. Thus our Edinburgh office could have purchased London out-of-area
exchangelines from the PTO. In effect, thismakesthe office appear to the public
telephone user as if it were in London. Itwill have a London telephone number and call
charges will assume that London is the endpoint. Thus calls from the Edinburgh to
London office, or vice versa, count as 'local calls'. There is a rental charge, however, for
the out of area line, similar to a leased circuit charge for a London/Edinburgh circuit.
- 828 CORPORATE
45.7 AWORDOF WARNING
Havingdescribedanumber of clever routing methodologies to reduce the costs of
private networks, it is worth recalling, before we end the discussion, that the end-to-end
connection must be properly serviceable. The transmission planning and routing ‘rules’,
set out in Chapters 28 and 33, are applicable here. For example, it is no use creating a
super-cheap end-to-end telephone connection which comprises so many component
links that speech is unintelligibleoverit.However, the high performance of digital
switching and transmission makes this much less likely.
If the private network is to be connected to the public network there may be legal
restrictions on the type and technical characteristicsof the signal that may be conveyed.
In some countries telephone network bypass is illegal. In nearly all countries there will
be signal power and technical line limitations.
45.8 PTO LEASED CIRCUIT OFFERINGS
Generally, PTOs offer a comprehensive range of different leased (leaselines) or private
circuit types,each attunedto one oranumber of particulartypes,suchasdata,
telephone or packetswitching.Withcare,theplanning of privatenetworks can be
relatively straightforward.
The five most common types of leased circuit conform to ITU-T recommendations
M. 1020, or M. 1025, M. 1040, their national equivalents, or they are digital circuits.
leased
Recommendation M.1020 defines the conditioning line-up
line (i.e. limits, see
Chapter 36) of a high grade four-wire analogue circuit, of 3.1 kHz bandwidth, suitable
for high speed data orspeech use. A circuit lined up according to M.1020 has a very flat
frequency response (i.e. shows very little attenuation distortion), very good group delay
response, and good circuit stability. Such a circuit is suitable for most of the analogue
data modems described in Chapter 9 and can be used for point-to-point or networked
(e.g. packet network) applications.
Recommendation M.1040 defines the line conditioning of a four-wire leased circuit,
intended primarily for voice conversation. Its performance is not as good as that of an
M.1020 circuit, but because time and equipment is needed for line conditioning, the
less
cost to the customer is lower. M.1025 provides a quality midway between M.1040 and
M.1020.
Digital leased circuits are usually defined in terms of their bit rate (e.g. 64 kbit/s,
2 Mbit/s (El), 1.5 Mbit/s (Tl), 34Mbit/s(E3), 45 Mbit/s (T3), 52Mbit/s (STS-1) or
155 Mbit/s (STM-l)), their biterrorratio ( B E R ) (typically 10-5 or 10-9) and their
overall availability (typicallyaround 99.5%). They are usually presented with oneof the
followinginterfaces:V.24(RS232),X.21V.35,V.36(RS449),G.703 or an ISDN-like
interface. Provided that these parameters suit the given application, the circuit may be
used for any type of service (e.g. telephone, data).
Other types of leased circuits that are sometimes available are wideband analogue
(e.g. FDM group, supergroup) or high speed digital circuits. In addition some PTOs are
willing toprovidejustthe physicalmedium, and allowtheuser to configurethe
bandwidthinthemanner desired.Examplesinclude the wholesale of opticalfibre
- MAKING USE OF MOBILE RADIO TECHNOLOGY 829
capacity (so-called dark fibre) and the leasing and sale of satellite dishes for direct
customer access to very high bandwidth satellite circuits (e.g. International Business
Service ( I B S ) , which is a satellite service of INTELSAT allowing companies direct
access for rates up to 2 Mbit/s). Finally, other specialized leased circuit services may be
available from particular PTOs.
45.9 MAKING USE OF MOBILE RADIOTECHNOLOGY
The companiesthat propose to mobile communications must careful attention to
use give
the coordination and managementthe system. Proper control must established over
of be
0 equipment
purchasing
0 air-time
subscription
0 cost
management
0 user-justification
rules
Urgency of
contactneeded
-
Radio pager Cellular phone
recommended recommended
High @ Company chairman
@ Hospital doctor @ Salesman
-
Ordinaryfixed Telepoint cordless
telephone recommended handset ( C T 2 )
recommended
Low @ Office
typist
@ Gordener
Low High
Mobility
( W i t h i n 5 minutes (More thon 5 minutes
of telephone) from telephone)
Figure 45.7 ‘Quantification’ chart for mobile communication users
- 830 CORPORATE NETWORKS
By dealing with a single supplier for carphone and hand-held units, and also for ‘air-
time’ supply, the administration and control of costs is made considerably easier, but
the ‘buying-power’ leverage over the supplier also increased. The decision to purchase
is
cellular or other mobiletelephones for useby individual departments needs to be
tempered by a central coordinating unit. One of the problems is that the falling cost of
handsets is bringing many of them withinthe financial authorization of the lowest levels
of management.
To get around the problem of integrating new and old generation equipment in this
fast-changing environment one tactic could be to rent the handsets. However, because
much of the involved
cost is made up of usagecharges(subscription and call
charges)usersshould not betempted toattachtoomuchimportancetowhatthe
secondary issue of handset cost. Usage costs of cellular radio technology are typically
around &25 per month per handset, plus a per minute rate for calls of up to three
timesthe ordinary telephonycharges.Cordlesstechnology(e.g. DECT) is slightly
cheaper but nonetheless can represent a substantial extra cost over normal telephones
or rudiopugers. For this reason, companies are recommended to employ some means of
assessing their users’ needfor service. A simple technique is shown diagrammatically in
Figure 45.7.
Figure 45.7 classifies users into four broad categories according to their ‘mobility’
(how long they are likely to be away from the telephone) and their ‘need for urgent
communication’ (i.e. whether they are likely to need to contact or be contacted by
others urgently). Examples of typical users in each the fourcategories of mobile com-
of
A
munications are shown. hospital doctorwith low mobility but high urgencyof contact
is recommended to have a radiopager, whereas for a company chairman or salesman
with high mobility and high contact urgency, a cellular telephone is recommended.
However, if this service is heavily congested a public cordless network (e.g. DECT if
available) might provide some degree of alternative. For an office typist we recommend
no mobile telephone at all, and our recommendation for the DECT handset is the
humble gardener, assuming of course he does not intend to‘lie low’.
nguon tai.lieu . vn