Some analysts were skeptical that Newbridge will do much to
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ improve Alcatel's business, because the industry is moving to
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ networks that run on Internet-based technology.
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ ``Newbridge would give them a great position in the ATM
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ market, but it's a short-term fix,' said analyst Eric Burkel of
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ Handelsbanken Markets.ÿ
Poor guy, does not realize that IP and ATM protocol are both packet and IP services can be run over ATM but not reverse.
He is also out of reality and does not see what SBC,Bell South, Bell Atlantic and others carrier in EU are doing on massive scale.
Both IP and ATM will be here for many uears to come.
The IP Syndrome
comsoc.org
Note from the Editor-in-Chief
My Message from the Editor-in-Chief devoted to the future of telecommunications
networking that appeared in the June
1999 issue of IEEE Communications Magazine has triggered interesting and fruitful
discussion. As expected, there is no single
opinion on how networking technology will evolve. I do believe that our magazine is a
perfect forum to exchange ideas as well
as to present different opinions, even controversial ones. Below, please find one such
opinion. I welcome other points of view
related to it as well as other comments on trends and developments in the communications
area. You can contact me at
a.jajszczyk@ieee.org
Andrzej Jajszczyk
This is a modified version of an article that appeared in Przeglad Telekomunikacyjny
(Telecommunication Review) in
January 2000. It is republished here with permission.
Consider the following rhetoric: The development of the information infrastructure leads to
the development of an
information society, which will be a next major step in the evolution of civilization. The
Internet has demonstrated the way the
information infrastructure can and should evolve. Thus, the Internet is the driving force
toward a new era in civilization.
This kind of extremely simplistic reasoning is being repeated over and over again by mass
media, businesses, and
politicians. In effect, the Internet is becoming a fetish rather than just a technological
solution; the center of a new ideology. In
the context of information and communication technology the new ideology manifests
itself with symptoms the author has
chosen to refer to as "IP-ism" or the "IP syndrome" (although not literally in the
psychiatric sense). In short, IP technology (its
derivatives, modifications, etc.) is regarded as a panacea for the development of a future
global and universal information
infrastructure ("everything over IP, IP over everything"). Let us take a closer look at
some symptoms and potential
implications of the IP syndrome.
The regulated and technology-driven development pattern of telecommunications has
rightly been criticized: the
disappointing pace of the dissemination of integrates services digital network (ISDN) and
broadband ISDN (B-ISDN)
services/applications, despite considerable conceptual and technological achievements,
resulted to a large extent from
neglecting the demand for services and applications -- the focal point was "how" instead
of "for what" and "for whom." The
Internet entered the underdeveloped space of services and applications so effectively and
with such an impetus that it is quite
hard to escape the impression that, indeed, a king's road toward a universal information
infrastructure was found. The birth of
IP-ism is thus understandable.
Faith in IP-ism is strengthened with the following argument. Data traffic is growing at
such a rate that, soon, voice
traffic will become relatively so small that it will be handled, with negligible effort and cost,
by the IP network in the
background of data traffic. Since data traffic growth is currently associated mainly with
Internet applications, it is obvious that
the means of information transport developed in the context of the public switched
telephone network (PSTN)/ISDN/B-ISDN
should be dropped and replaced by transport based on IP. This argument is logical, but
only on the condition that contrasting
data traffic with voice traffic is assumed reasonable.
This, however, is the case as long as the term data refers to information flows associated
with services/applications
similar in nature to those known from today's Internet, which are mostly
man–machine-type interactions. Entertainment,
education, e-commerce, and many other applications will sooner or later demand
"humanized" multimedia communication
resembling face-to-face presence (even if only in virtual reality) rather than just
"user-friendly" interaction with smart
machines. Such communication requires real-time high reliability and quality exchange of
visual, audio, and symbolic
information. The visual component (dynamically changing images, video, etc.) will likely
dominate the total multimedia traffic in
terms of bit rate. Referring to this type of traffic as data does not make much sense since
in such a case practically any kind
of traffic would be called data. Moreover, the term data becomes potentially misleading
because of its historical association
with applications of computer networks. This association apparently leads many to the
conclusion that computer network
technology, and particularly the IP technology, is an obvious solution for the future
universal information infrastructure. This
conclusion, however, is based on:
Misinterpretation of the term data
Oversimplified extrapolation of the current traffic growth (assumed domination of data
traffic associated with
man-machine type applications) together with underestimation of the future importance of
humanized communication
Superficial technological considerations (computer network technology considered
universal)
Considering that the application of IP technology to "humanized" multimedia
communications faces "unexpected"
problems (real-time communication regime, quality of service, etc.) and that these
problems are unlikely to be solved soon (see
the following remarks), it is quite probable that the development of services/applications in
the near future will be tailored to the
current abilities of IP technology. If so, the development of the information infrastructure
will become technology-driven rather
than services/applications-driven (i.e. it will follow the development patterns which were
rightly criticized). Paradoxically,
market forces are enforcing this trend since with the current tremendous success of
Internet it is risky for service and network
providers to bet on any other technology in the near future. And what about the more
distant future? Well, the current
development of the market is so fast that strategic planning is reduced to the perspective
of single years; a more distant future
"does not exist."
Leaving aside market and psychological phenomena, let us focus on technological aspects
of IP-ism. The communication
functionality of IP-based networks can, of course, be enhanced and perfected (with
today's technology, "anything can be
done") in order to remove its current well-known limitations. The task, however, is not an
easy one; the problems associated
with IPv6, DiffServ, IntServ, and so on demonstrate this quite clearly. It is also becoming
clear that the problems to be solved
are in essence the same as encountered by the developers of the B-ISDN/asynchronous
transfer mode (ATM) networks.
After all, IP communication technology is just a form of packet switching, and as such
does not provide any miraculous
solutions to tasks that are inherently involved in developing communications services in an
healthy social and market
environment: mass scale, affordable cost, appropriate QoS, service and network
management, billing, and so on. Plus ça
change, plus c'est la méme chose.
The complexity, generally speaking, of managing the development of services and
networks is growing rapidly when new
customized services and applications are introduced. Coping with management issues is
likely to be the hardest part in
developing the future information infrastructure. The telecom engineering community has
successfully dealt with such tasks,
but only in a comfortable plain old telephone service (POTS) monoculture, and thus
lessons from the past are of limited
usefulness. On the other hand, the Internet engineering community has just recently faced
such tasks. Apparently, both
engineering communities are becoming aware of the challenge and likely will soon meet
on common grounds.
Considering the above, the choice of the most appropriate switching, multiplexing, and
communication mechanisms for
the future information infrastructure should be subordinated to problems associated with
managing
networks/services/applications, QoS, billing, and so on, and not the other way around.
IP-ism seems to adopt the latter
approach. The success of the Internet in delivering useful and fascinating new
services/applications apparently has caused an
impression that all problems can be solved if patterns that have led to success in the past
are followed in the future. One of
these patterns was the separation of the communication (transport) functionality from the
service/application functionality. This
enabled the creation of services/applications (on the edge of the network) without
changing the transport mechanisms (i.e.,
keeping it simple). In contrast, introducing new services/applications in
PSTN/ISDN/B-ISDN is associated with the
modification of network communication (with the exception, to some extent, of intelligent
network services), which is a tedious
task due to the scale and complexity of the network. In effect, introducing new services or
applications becomes difficult and
time consuming since it requires standardization, long-term and centralized planning, large
investments, and so forth.
The Internet approach works well as long as the transport network is neutral to
services/applications -- "transparent"
(i.e., has virtually infinite capacity) -- or no specific requirements concerning its features
as an information transport medium
are posed (i.e., best effort transport is assumed satisfactory). The first possibility is
unrealistic and the second unsatisfactory
(in the long run); thus, obviously a compromise between the Internet and the
PSTN/ISDN/B-ISDN approaches is required.
The network communication functionality cannot be too simple and decoupled from
services/application because this limits
control over performance (QoS, manageability, etc.); on the other hand, complex network
functionality closely coupled with
services/applications makes the whole system inflexible in terms of development
(introduction of new advanced
services/applications).
Believers in IP have no better solutions to the problems indicated above than do, say,
believers in ATM. The IP and
ATM network concepts were born in technological, service/application, and market
conditions which differed substantially
from those of today. Both IP and ATM can, of course, be improved and modified, but
there is no evidence that this will result
in future-proof solutions. Sticking to IP and ATM technologies might lead to eclectic
solutions -- overloading networks with
features to prolong the life of technologies which will come to an end quite soon anyway.
Adding complexity to IP networks
goes against one of the original sources of the Internet's success; keeping stiff ATM
formats is no longer justified in today's
state of the art in switching and routing technology.
IP-ism leads to ideas like IP over dense wavelength-division multiplexing (DWDM), with
the omission of synchronous
digital hierarchy (SDH) and ATM systems. The idea is tempting since it eliminates two
network layers and leverages the
troublesome interworking of SDH, ATM, and IP. If realized, however, some important
network functions performed by SDH
and ATM will have to be provided within IP networks; in general, it is the decomposition
of high-capacity channels into
hierarchically ordered subchannels. It is an illusion that large networks can be built without
well defined decomposition
principles and techniques. Constructing large and complex networks that are effectively
managed in terms of quality, reliability,
functionality, profitability, growth, and so on is first an art of structural and functional
decomposition. IP over DWDM
networks risk lacking structural and functional clarity ("all in one") if the development of
the idea is not guided by a reference
architecture for network resources and capacity management. So far such an architecture
has not been established; instead,
IP network developers seem to believe that all problems can be handled with a smart
communication protocol (e.g., DiffServ,
IntServ).
Supporters of all-IP network solutions often refer to the high traffic handling effectiveness
and scalability of IP networks
resulting from asynchronous multiplexing of traffic streams and the connectionless
communication mode. High traffic handling
effectiveness and scalability, however, are achieved "easily" only if the best effort
principle is maintained. Introducing
QoS-controlled communication functionality to best effort networks requires much more
than just facelifts or bypasses.
Despite a considerable effort made to solve the problem with sophisticated communication
protocols, there is still no evidence
that the envisioned solutions will actually work in large-scale multiservice networks.
In principle there is a trade-off between the complexity of traffic handling (effectiveness,
QoS issues, etc.) and the
available network capacity: the more capacity can be provided, the less complex
communication functionality is required to
cope with traffic handling problems. In the current technological and market
circumstances, available capacity for information
transport in the core network grows faster than the information processing power
available in network nodes. This tendency
will probably last for some time. It thus seems reasonable to trade network node
complexity for network transport capacity:
keep network communication protocols simple at the expense of network transport
capacity. This implies that it might be more
reasonable to "waste" capacity in order to achieve manageability by means of structuring
network resources (hierarchization of
network nodes, layering of transmission channels) rather that through increasing
complexity of network communication
protocols in a flat network structure. Choosing the right balance is not an easy task: how
much capacity can we afford to
"waste," considering that any available capacity is usually rapidly exhausted with our
appetite for new services and
applications? Are SDH and ATM network structuring means appropriate, or should we
look for other means? The issue
deserves much more attention than it has been given so far.
Controlling complexity becomes an issue per se. Functional complexity of technical
systems, especially in the information
and communications technology area, has grown several orders of magnitude during the
last two decades (compare, e.g., the
volume and complexity of standards in the communication area in the '70s and in the '90s).
There seems to exist a widespread
conviction that any degree of complexity can eventually be mastered with digital
technology. Nevertheless, with digital
technology we have achieved greater perfection in controlling a spaceship than a broom
(robots are still clumsy in performing
some tasks which are simple for humans and even for animals). Managing an information
infrastructure -- if it is to serve
people! -- is not only a matter of putting another digital system on top of it. The belief that
digital technology is a remedy is not
well grounded. In the near future the complexity of digital systems might become as
troublesome as industrial pollution today.
In spite of the problems indicated here, IP-ism is flourishing. The unquestionable success
of the Internet has raised such
emotions and hopes for the future that even slight skepticism concerning the technological
and economic rationality of IP-ism is
considered defeatism. Obvious problems with introducing, say, voice over IP (VoIP) with
reasonable quality on a large scale
do not seem to disturb the enthusiasm. The stock value of IP-related businesses is a good
measure of the optimism.
Neglected problems do not disappear, but rather accumulate. An optimistic scenario is that
the problems will soon trigger
the development of a new technology which will draw from the best features of IP, ATM,
TDM, and so on. The pessimistic
scenario is that IP-ism will stay influential long enough to make us believe that services
and applications IP technology is able
to provide, even if they do not match our expectations, are "objectively" the best possible
in current circumstances. Such a
scenario is not improbable since the Copernicus-Gresham law apparently also applies to
the communication services market;
poor-quality voice services (VoIP) might push out traditional telephony (hopefully, "voice
over IP" will not lead to "IP instead
of voice"). The author would be more optimistic i,f along ,with introducing VoIP the
historical telephony standard (300–3400
Hz !) was finally enhanced |
