chapter39

Flexibility
The TDM PBX is a proprietary and inflexible device, closed in every respect except
for its CTI interface, which provides for limited call control. Although IP PBXs
(with the exception of open-source systems) do not open their call-control programs,
they provide more flexible interfaces such as SIP to permit development
of server-based features. They also open options for branch and home offices that
are expensive or difficult with traditional architectures.
Create Value-Added Services
Convergence opens an enormous variety of opportunities to provide new services
that are infeasible with circuit switching. The user interface for telephone service
has improved little over the years. This is not because of a lack of imagination on
how to improve it or a lack of APIs for hooking new applications to proprietary
systems. It is more because the applications must be customized for each type of
CPE system, and no manufacturer has enough market share to generate a mass
market. For example, unified messaging has been available for years, and there is
little variance in the features that various products support, but it has not achieved
enough market penetration to bring the cost down to the point of becoming popular.
Open protocols such as SIP can separate the services from the call control,
which by its nature must be closed. Carriers and third-party developers can create
countless new services and make them operable across a variety of platforms.
Enriched User Experience
Ultimately, new services will change the way people work and communicate.
Personal communication assistants can enable critical employees to be contacted
while still screening unwanted calls without the need for human assistance.
Functions that are difficult with a standard telephone interface, such as setting
up conference calls and dialing by directory name, will become easier with an
improved user interface. Productivity should improve through remote collaboration
and shared access to documents or whiteboard. Just as the PC is a standard
office tool today, these new applications will become such a way of office life that
users will expect them to be available.
Rapid Deployment of New Applications
New and innovative applications can be deployed more rapidly with IP than
with traditional fixed telephone systems, and the pace of improvement does not
depend on the actions of a single vendor. Furthermore, in a geographically dispersed
organization, new applications can be downloaded onto desktop clients
without involving a generic program upgrade.
Barriers to Convergence
New telecommunications developments have always been over-hyped, and with
notable exceptions such as fiber optics, many of these have flared briefly and fizzled
out. After an initial period of exuberance, convergence is proceeding slowly for a
variety of reasons, not the least of which is the difficulty in demonstrating a suitable
return on the investment. Most of the advantages listed above require the
organization to change and adapt to a new environment, and this often happens
slowly.
The initial impetus was long distance cost saving, but that argument has
largely disappeared now that long distance costs are so low. Savings from managing
the converged network are difficult to prove unless the workforce shrinks
with the new technology. When ROI is calculated, many of the savings are in soft
dollars and demonstrating real cash saving is more difficult. Convergence will
develop in time, but many issues listed in this section remain to be resolved.
Network Infrastructure
Carriers are converting portions of their network to VoIP, but islands of VoIP
cannot support the performance and security that commercial-grade voice
communication demands. For VoIP to be a viable alternative to the PSTN, it must
support carrier interoperability. For services such as worldwide VPNs and telephone
connections to be effective, they must transcend carrier boundaries because
no carrier can fulfill all of the needs without relying on other providers. Carriers
must be able to provide services to any corner of any rural area of any country,
and no carrier has sufficient reach without relying on other service providers.
Carriers therefore must be able to interconnect with appropriate levels of security
and service definition and the sessions must be metered to compensate carriers for
handling transit traffic.
Today, many outsiders expect the Internet to become the backbone for this
multiservice, multinational, multiowner network, but those expectations are unrealistic.
Anetwork converged over IP does not mean the public Internet as it is now
structured. Obviously, it can carry voice. It does it every day, but it cannot carry
voice with the consistent quality that the world has enjoyed since the conversion
to an all-digital network. The alternatives are developing an overlay network that
has the stability isochronous applications need, or hardening the Internet. The
latter means changing the basic design concepts that keep the Internet cheap
and fast-paced.
Peering points on the Internet today do not meet any specified performance
criteria and there is no incentive to support a guaranteed level of service. The
converged network must provide an appropriate level of assured delivery in
response to requests from the application. This is inconsistent with the intent of
the Internet, which is to deliver inexpensive connections that are not sensitive to
usage or distance. Either the Internet must be split to provide a separate network
with reliability and security, or the cost of service must increase.
Flow Control
Another key issue is congestion control, which is a vital feature of any voice or
data network. The difficulty is that voice and data behave differently when it
comes to congestion. Both can throttle traffic back at the source, but the nature of
the traffic flow is much different. Many data applications have peaks of high
bandwidth demand for short intervals, but then demand drops to zero as the user
operates on a downloaded file. If the network is congested, it is apt to be for only
a brief interval, after which traffic begins to flow normally. During the heavy flow
periods, TCP closes its window or routers discard traffic, but the process is transparent
to the users, who see a slow response, but the session continues without
interruption.
Unlike data with its heavy peaks, voice is a relatively even flow of half-duplex
traffic that is predictable. Traffic engineers have mounds of data that enable them
to predict voice loads by hour, day, and season until something unusual happens.
Storms, disasters, significant news events, and other external events usually
inspire an extraordinary number of customers to place telephone calls. These
cause traffic to fall outside the normal range and the network has to protect itself
while prioritizing service to essential customers. Voice networks shed load by a
variety of techniques, the first of which is to delay dial tone. During heavy load
periods the LEC can operate line-load control, but this is done only in extreme
circumstances. Common-control equipment such as DTMF registers are engineered
for normal peaks. In abnormal peaks, the registers may be tied up, so the
caller does not receive dial tone. The caller can remain off the hook and dial tone
will eventually be provided. If the congestion is in the trunking network, calls will
not go outside the serving class 5 switch. The user hears reorder and must redial.
Flow control is a standard feature of TCP, but real-time packets work under
UDP, which does not provide flow control. An IETF working group is working
on Datagram Congestion Control Protocol, which is intended as an alternate
transport protocol. DCCP offers functions that bridge the gap between TCP and
UDP. These include packet acknowledgement, congestion notification and
control, packet sequencing, and protection against denial-of-service attacks. This
protocol may resolve flow control issues.
Security
This is the issue that is the most difficult to resolve on the Internet, while still
keeping the service manageable. While users must get involved with encryption,
tunneling, firewalls, and similar security provisions, VoIP will be confined to a
narrow spectrum of users that are willing to put up with the complexity in
exchange for the benefits. A major strength of the PSTN is the fact that uninvited
guests cannot ride the coattails of a file or message, latch onto the telephone, and
infect it with a virus. While trust is not an issue with the PSTN, eternal vigilance
is required on the Internet to thwart a coterie of miscreants who, for whatever
malevolent motivation are attempting to inflict damage.
Broadband Penetration
For VoIP to be successful, broadband must become nearly as ubiquitous as the
PSTN. In 2004, the latest figures available as this book goes to press, broadband
penetration in the U.S. is reported to be 42.5 percent of households with Internet
access. About three-fourths of households have Internet access, which means
broadband penetration is roughly one-third of the households. Other countries,
notably Sweden, Japan, and South Korea, have penetration in the order of threefourths
of all households, significantly greater bandwidth, and at a cost that is
more affordable than in North America. This issue must be resolved before the
converged network can claim success.
Service Complexity
For VoIP to become universally accepted, it must be simplified. Today, customers
need to know too much about VoIP to make it work. The strength of the PSTN is
its simplicity. Customers do not need to know anything about addresses, E-911
access, NAT, and other such technical issues to make the telephone work. For VoIP
to be ready for real time, users must be able to plug the service into a wall jack and
have it work without the need to configure firewalls, install VoIP terminal
adapters, or worry about IP addresses. Users must be able to connect to other VoIP
users without using the PSTN as an intermediary and without the need to know
the identity of the callee’s service provider.