The Path to
Convergence
by Jim Metzler
The goal of this paper is to reduce the gap between the enthusiasm of the
networking vendors and trade publications for the topic of convergence and the
reluctance on the part of enterprise users to implement it.
As a first step towards that goal, we must agree as to what is meant by
convergence. To the author, a converged infrastructure is comprised of one or
more of the following components:
Converged Applications
Deploying applications that use a combination of voice, data, and video.
Converged Network
Combining one or more forms of traffic onto a single LAN and/or WAN. The traffic
types include voice, different types of data, and video.
Converged Management
Combining the tools, processes, and people that currently manage disparate
pieces of the IT infrastructure into a common, efficient system.
Perhaps the most discussed aspect of convergence is IP Telephony. The author
uses the term IP Telephony to mean providing voice service over an IP network in
a manner such that the users of the service have virtually the same experience
that they would have had if the voice had been transported over a circuit
switched network. However, in order to accomplish this, the network must provide
low, predictable delay as well as the key features of a voice service that users
have come to expect.
For convergence to be successful in the marketplace, it must add business value.
There are many cases where this is possible. For example, many companies want to
create tighter relationships with customers, suppliers, and distributors. It is
now possible to provide a single interface to the customer (converged network)
for all types of interaction, while providing customer care applications that
combine voice and data (converged applications) and Policy Management (converged
management) to ensure that applications get the resources they need.
Two key components of managing a converged infrastructure are policy management
and service level management. The author uses the phrase "policy
management" to refer to the ability to dynamically configure traffic flows
based on user, application, or outside event. As defined, policy management has
both a tactical and a strategic impact. On a tactical basis, policy management
allows IT professionals to both automate and apply intelligence to configuring
the IT infrastructure. On a strategic basis, policy management enables:
_ Directory
Integration,
_ Quality of
Service, and
_ Security,
end to end across the network.
A converged WAN consists of either converged WAN access or a converged WAN
backbone, and it logically extends the concept of IP Telephony to include
multiple applications on a common data-centric WAN. The phrase "converged
WAN access" refers to the dynamic combination of voice, video, and data
traffic onto a single transmission group from an enterprise location to the
carrier's Point of Presence (POP). To understand how converged WAN access can
help to reduce costs, consider a company that has 200 branch offices with
multiple T1s from each branch office to the carrier's POP. Assume that the
average monthly cost of these T1s is $500 and they are statically configured to
carry voice, SNA, and IP traffic. Further assume that by deploying Inverse
Multiplexing over ATM (IMA), the enterprise reduces one T1 per office. This
results in a $100K monthly savings, or $3.6 million over a three-year life
cycle.
A converged WAN backbone can help an enterprise both to save money and to
compete better in the marketplace. Nortel Networks is a good example of a
company who has achieved these two goals by deploying an ATM wide area network
to carry voice, data, and video traffic. Five years ago, the company had seven
separate but overlapping networks serving the voice, data, fax, and video needs
of roughly 60,000 users in both Nortel Networks and Bell Northern Research
locations worldwide. Nortel Networks deployed a network based on
carrier-provided ATM services in the U.S. and T3s both to and within Canada. The
network carries all of Nortel Networks' traffic and had a payback period of less
than a year. In addition, the network has reduced Nortel Networks' time to
implement network upgrades and has scaled in a manner that most likely would not
have been possible with the prior disparate WAN infrastructure.
As we enter the next millennium, the competitive pressures on businesses will
only increase. In addition, over the next several years, the success or failure
of a business will increasingly be linked to the success or failure of the IT
organization. Given these pressures, it is critical that the IT organization
expand their efforts to align the goals of the IT organization with those of the
business. What is so appealing about the deployment of a converged
infrastructure is that the benefits it offers match up so well with the most
common issues that currently impact business. However, to realize the benefits
of convergence, today IT organizations need to deploy the infrastructure that
can support it. They also need to develop and implement a plan to ensure that
they have the skills in the organization to interface with the Business Units on
a more strategic basis, as well as to implement and manage the new technologies.
1.0 The Myths and Realities of Convergence
It is hard, if not impossible, to find a networking vendor who is not strongly
advocating the deployment of convergence. It is equally difficult to read an
industry publication and not find one or more articles discussing the advantages
of convergence. However, when asked, most enterprise IT organizations state that
they are notably less committed to convergence than either the networking
vendors or the industry publications. The purpose of this paper is to reduce the
gap between the enthusiasm of the networking vendors and trade publications and
the reluctance on the part of enterprise users.
In order to accomplish this, the paper will synthesize some of the
existing documentation relative to convergence, crisply identify the benefits of
convergence, discuss the key challenges relative to managing in a converged
environment, provide examples of how Nortel Networks has begun to deploy
convergence within their internal network, and suggest some next steps for those
enterprise IT organizations looking to evaluate convergence.
Enterprise organizations that are evaluating whether or not convergence makes
sense for them face a number of hurdles. One such hurdle is that there is not a
common definition in the industry as to what convergence means. A second hurdle
is that there are several myths surrounding the benefits of convergence as well
as the difficulty of deploying it.
The purpose of defining the term convergence is to be able to succinctly and
accurately debate the value of convergence to the enterprise. One possible way
to define convergence would be very broad. Using this approach, the definition
would include things like the industry's convergence onto Ethernet as the LAN
technology of choice for the desktop, or the industry's convergence onto TCP/IP
as the protocol of choice. However, since both of these phenomena have already
occurred, including them in the definition of convergence will not facilitate
the debate over the viability of convergence on a going forward basis.
To facilitate this debate, the definition of convergence needs to include
components that are not already realized in the marketplace but which offer
significant promise of being realized. Towards that end, the author suggests
that a converged infrastructure is comprised of one or more of the following
components:
Converged Applications
Deploying applications that use a combination of voice, data, and video.
Converged Network
Combining multiple forms of traffic onto a single LAN and/or WAN. The traffic
types include voice, different types of data (i.e., email and ERP), and video.
This can also refer to combining multiple traffic types onto a single circuit
between an enterprise's offices and their carrier's Points of Presence. The most
talked about instance of this type of convergence is moving voice, fax, and
video applications over an IP infrastructure.
Converged Management
Combining the tools, processes, and people that currently manage disparate
pieces of the IT infrastructure into a common, efficient system.
As previously mentioned, there are several myths surrounding convergence. The
following list of myths is based on a Lucent document.
Myth: Convergence means that everything a company has today will soon be
obsolete.
Reality: Convergence is both an evolution and a revolution. Many parts of
today's networks will continue to deliver value for years. In addition, there
are instances in which separate networks will continue to exist.
Myth: Convergence means just running voice over a legacy data network.
Reality: To be successful, converged networks must be able to deliver virtually
the same quality of voice services that customers have come to expect. However,
the current lack of reliability, manageability, and security of most data
networks precludes the broad deployment of voice over these legacy data
networks. In addition, as voice applications migrate to a converged environment,
customers expect the same call billing services they have today in their voice
environment. This level of billing is virtually non-existent in a data network.
Myth: Convergence is primarily about cutting costs.
Reality: Convergence offers the potential to cut costs as well as to provide a
competitive advantage to the enterprise. This competitive advantage typically is
a result of the new applications either enabled or made notably easier by a
converged infrastructure.
Myth: Ultimately there is one right path to convergence.
Reality: There are multiple paths to convergence based on where the enterprise
is starting and what is driving it. However, as will be discussed in section 4
of this paper, any path to convergence has to include deploying the capability
to effectively manage the converged infrastructure.
Myth: Convergence is all about increasing the bandwidth.
Reality: Most corporations will not want to just throw bandwidth at the problem
- at least not in the WAN. As a minimum, what will be needed is a way to
allocate the bandwidth appropriately and effectively to ensure that certain
applications such as voice receive low predictable latency.
2.0 Drivers of Convergence
Our industry has a long history of deploying technologies in search of a problem
to solve. Hence, in order to determine the viability of convergence in general,
and IP Telephony in particular, it is important to understand the degree to
which a converged infrastructure adds business value. To accomplish this, the
following lists some of the most common issues currently affecting businesses, a
statement as to how each issue translates into a requirement of the IT
organization, and a statement as to how convergence of applications, networks,
and management helps the IT organization to meet this requirement.
Business Issue: Reduce the time it takes for a company to bring products and
service to the marketplace.
Network Requirement: Reduce the time it takes for the IT organization to deploy
new services or to extend existing services to new sites.
Impact of Convergence: It is easier and faster to deploy new services over a
single robust infrastructure than over disparate infrastructures (see section
5).
Business Issue: Create tighter relationships with customers, suppliers, and
distributors.
Network Requirement: Make it easier for customers, suppliers, and distributors
to do business with the enterprise.
Impact of Convergence: It is now possible to provide a single interface to the
customer for all types of interaction, while enhancing the experience by
providing customer care applications that combine voice and data traffic.
Business Issue: Reduce operating costs.
Network Requirement: Reduce the unit cost of running a network.
Impact of Convergence: The people, transmission, and capital costs of a
converged infrastructure are less than those of multiple disparate
infrastructures (see section 5).
Business Issue: Attract and retain the best talent.
Network Requirement: Extend the network to homes and small offices.
Impact of Convergence: It is now possible to provide employees with one place to
go for all of their communications; i.e., voice, fax, data.
Business Issue: Implement an eBusiness strategy.
Network Requirement: Increase the performance of certain applications.
Impact of Convergence: As discussed below, implementing Policy Management
facilitates providing the performance that a given application needs while
providing the remaining applications sufficient resources.
Business Issue: Improve productivity.
Network Requirement: Support applications such as white boarding and video
conferencing.
Impact of Convergence: It is possible to deploy productivity tools that combine
voice, data, and video, and via Policy Management ensure that these applications
get the resources they need.
Business Issue: Increasing revenues.
Network Requirement: Support increasing traffic volumes.
Impact of Convergence: It is more efficient to support ever-increasing traffic
on one relatively large network than on separate, relatively small networks (see
section 5).
One of the first applications of a converged infrastructure is IP Telephony.
Note that the author uses the term IP Telephony to mean providing voice service
over an IP network in a manner such that the users of the service have virtually
the same experience they would have had if the voice had been transported over a
circuit switched network. In order to provide for IP Telephony, a converged
network must be able to guarantee end-to-end delay of roughly 150 ms or less,
and that the delay will not vary by more than 20 or 30 ms. However, in addition
to providing low, predictable delay, the network must also provide the key
features of a voice service that users have come to expect.
Focusing on IP Telephony, there are two key factors that typically drive an
enterprise to deploy it. The first is cost savings. In particular, over time
packet switching is becoming notably more cost-effective than circuit switching.
This fact was documented in a 1997 article in Business Communications Review
(BCR) written by Peter Sevcik of Northeast Consulting. What that BCR article
shows is that packet switching will soon be between twenty and fifty times more
cost-effective than circuit switching.
The second factor that drives enterprise users to IP Telephony is the ability to
deploy new capabilities that improve user productivity. One such capability is
the ability to browse Web pages from a phone. Underlying this capability is a
technology referred to as VXML (Voice eXtensible Markup Language). VXML is
somewhat similar to HTML. When a user calls a special phone number, the call is
routed to a device called a voice response unit. This launches a Web browser,
which finds and interprets a document written in VXML and then responds to the
caller. Users can interact with the Web either by voice or touch-tone.
Other capabilities that convergence either enables or makes notably simpler
include:
Virtual Call Centers
Due to a shortage of skilled labor, many corporations are forced to extend their
call centers to where they can best hire agents. A converged network gives the
corporation the ability to easily implement a virtual call center in which the
network functions like the traditional call center switch.
Simplification of Routing Administration
Voice network routing table administration can be significantly simplified by
integrating the voice network routing table with the distributed, autonomous
routing intelligence inherent in packet-switched networks. The flexibility of IP
networks can significantly simplify moves, adds, and changes within telephony
networks.
Mobility
IP Networks offer the ability to associate names with users rather than with
physical ports on the network. This offers the flexibility to use the same phone
line for simultaneous voice mail, e-mail, and voice calling. This flexibility
can also be used effectively to allow employees to travel and receive calls at
their temporary locations.
Directory Integration
Data networks and voice networks have separate directories for their users. IP
Telephony enables the integration of multiple directories into a single cohesive
system. This eliminates both the labor and the errors associated with the
replication of information. Vendors that leverage their customers' investments
in corporate directories increase the productivity and viability of the overall
system.
Policy Integration
Traditional telephony enables the specialized treatment of calls based on the
user profile. Likewise, data networks have policies to control access to
information and manage network resources. IP Telephony allows the integration of
policies, based on a myriad of rules, across both telephony and data networks.
3.0 An Industry-Specific View
The previous section looked at some of the most common issues currently
affecting businesses in general and discussed how a converged infrastructure
would be of benefit. This section provides additional detail on three key
industry segments: the medical, retail, and financial industries. The section
will both describe the particular business pressures that are affecting those
industries and identify the degree to which these pressures can benefit from a
converged infrastructure.
3.1 The Medical Industry
The medical industry is undergoing a metamorphosis driven by a combination of
marketplace pressures, government regulations, and consumer demand. The medical
industry has moved from a time when physicians were in charge of medicine to one
in which the HMOs are in charge. In order to be more responsive, some HMOs are
trying to move to an environment in which patients are more in charge of their
treatment. Another key characteristic of the medical industry is the ongoing
mergers and acquisitions.
Because of these pressures, the IT organizations within the medical industry are
typically driven to:
_ Rationalize
the disparate networks that result from mergers and acquisitions;
_ Provide
consistent and improved customer service; and
_ Implement
enhanced security.
One of the particular factors driving the medical industry to need enhanced
security is the recently enacted government regulations that require electronic
signatures on any patient information that flows across a network. Medical
organizations that do not comply with these regulations face stiff fines.
Given this environment, most IT organizations within the medical industry are
not ready to make a major deployment of a converged network infrastructure.
Rather, this is the time for these organizations to begin to harmonize their
disparate networks to facilitate deploying a converged network infrastructure in
the near term. However, this is a very appropriate time for these IT
organizations to explore converged applications and enhanced network management.
In particular, tele-health applications can improve the quality of health care
while simultaneously reducing cost. The deployment of policy management (see
section 4.1) can greatly increase the security within these networks.
3.2 The Retail Industry
The retail industry is characterized by needing to support a high volume of
transactions, being very cost-sensitive, and needing to offer high levels of
customer service. These industry characteristics translate into the need for a
converged network infrastructure, converged applications, as well as for
converged management. In particular, the retail industry has the following
convergence requirements.
Wireless VoIP
Most large retailers are implementing wireless LANs and are moving towards
802.11 to support wireless scanning and computing devices. Many retailers would
like to use a common wireless infrastructure to support wireless voice.
The use of a common wired infrastructure
Retailers today run many different network infrastructures in a store. To reduce
operational costs, and potentially installation costs, it is highly desirable to
use a common wired infrastructure in new retail stores. One advantage of
Ethersets is that they would reduce the cost of moves, adds, and changes.
Another advantage of deploying Ethersets is that new functionality can be
introduced to telephones through software upgrades on a server. For example,
soft keys for calling the store manager, security, or building services can all
be on the phone. The server can be programmed remotely to ensure that the phones
will automatically use the correct numbers, even if they change often. In
addition, the store directory can be available with one-button access.
Reduction in telephone costs in each store
By using VoIP for store-to-store, store-to-distribution center, and
store-to-head office communication, retailers can potentially reduce long
distance charges and reduce their requirements for outgoing lines and trunks.
Enable the delivery of new services
One such service is to enable a retailer to transfer a telephone call received
in a customer care center to the appropriate store along with customer
relationship information on the originating caller.
Support other real-time traffic
Most retailers want to move their Point of Sale (POS) traffic to a common IP
network with the assurance that it will receive the highest priority and will
always get through.
Enable the use of common management infrastructure
By moving to all IP-enabled devices, a remote common management infrastructure
can be used. This tends to both reduce the cost and increase the quality.
3.3 The Financial Industry
The financial industry is undergoing a transformation due to factors such as
globalization and continuing deregulation. At the same time this is happening,
customers are becoming more sophisticated and demanding. This combination of
factors is creating pricing pressures throughout the financial industry.
These industry trends translate into a variety of key business objectives for
financial institutions. Financial institutions want the ability to:
_ Lower operating
and transaction costs
_ Increase
profitability per customer
_ Offer choice and
convenience to customers
_ Understand
the customer's whole needs
_ Provide
trained resources that can engage with clients
_ Maintain/increase brand equity and
extend trust
_ Implement a
flexible operating environment to stay ahead of competition
These industry characteristics translate into the need for a converged network
infrastructure, converged applications, as well as for converged management. For
example, one of the strategies that financial institutions have implemented to
respond to these business issues is to simplify and integrate both the
traditional branch office and the various other customer touch points. Disparate
networks, applications, and back-office systems make it very difficult to
service those touch points in a cost-effective, efficient manner. As a point of
reference, over 60 percent of a financial institution's IT costs are associated
with maintaining legacy systems.
Another key strategy for financial institutions to implement is the migration to
an IP environment, and to port applications to work in this environment. In this
way, the financial institution can begin to enhance the customer experience via
a Web interface as well as take advantage of the increasing number of
Web-friendly and multimedia applications designed to operate over IP. This now
positions the IT organization to bring voice and data together on the Web and
tie in the customer relationship management center (Call Centers).
The financial institution can now package and process their voice calls
and data records seamlessly. For
example, customer profile information and records need to travel securely from
the customer contact center to any other point in the enterprise and be
synchronized with the customer call. In this way the customer is managed and
treated consistent without information needing to be repeated. In order to
support this, an optimized IP network with high-performance QoS, reliability,
and security is essential. This drives the requirement for converged management
(see section 4.0).
4.0 The Role of Network Management
In a converged infrastructure, enhanced network management is both a requirement
and an opportunity. It is a requirement because a converged infrastructure will
require higher levels of reliability and security than legacy infrastructures
do. In addition, a converged infrastructure will also require that the network
be provisioned to provide low, predictable delay to those applications that are
delay-sensitive. A converged infrastructure also presents the opportunity to
provide enhanced network management. This opportunity is to manage the campus,
the WAN, and the telephony network as a single system. As previously mentioned,
deploying a converged infrastructure affords the enterprise IT organization the
ability to perform better routing and equipment administration as well as
integrate directories and policies across the enterprise.
Directory integration provides an effective solution to the disparate database
debacle. Specifically, directories
have historically been application-specific. Hence, the IT infrastructure is
littered with disparate directories for applications such as e-mail and remote
access. Existing methodologies for both updating directories and keeping
information consistent across them is both labor-intensive and error-prone. The
goal is to migrate to a single, unified directory that integrates the disparate
directories. This unified directory achieves this goal by adhering to a common
schema and interfacing with a policy server (see below) that enforces a set of
rules or policies.
4.1 Policy Management
Two key components of managing a converged infrastructure are policy management
and service level management. The author uses the phrase "policy
management" to refer to the ability to dynamically configure traffic flows
based on user, application, or outside event. The need to configure traffic
flows based on the user or application is an often-discussed topic. In
particular, it is generally accepted that it is desirous to treat people and
applications differently based on a variety of business criteria. However, the
need to dynamically configure the network based on an outside event is discussed
far less often. The issue here is that in many cases it is possible for an event
that is external to the extended enterprise to significantly change the traffic
flow requirements. For example, an uprising in some part of the world (i.e.,
Indonesia) would cause some of the users in companies such as brokerage firms
and insurance companies to need quick access to a variety of news broadcasts.
As defined above, policy management has both a tactical and a strategic impact.
On a tactical basis, policy management allows IT professionals to both automate
and apply intelligence to configuring the IT infrastructure. This capability
presents two significant advantages to the IT organization - reduced cost and
higher availability. Policy management reduces cost by easing the amount of
resources required to do configuration. Policy management increases availability
due to the fact that an outage in the IT infrastructure is as likely to be
caused by a configuration error as by a hardware error. As such, implementing
intelligence-based processes for configuration management can eliminate
configuration errors and the outages they cause. The strategic impact of policy
management stems from its ability to enable the enforcement of a set of rules or
policies that dictate access rights and use of resources to meet established
business objectives. For example, IT organizations are increasingly being
required to offer acceptable Quality of Service (QoS) to certain
mission-critical applications while also meeting the needs of other applications
and users in a cost-effective manner.
In general, the strategic impact of policy management comes from its ability to
enable:
_ Directory
Integration,
_ Quality of
Service, and
_ Security.
end to end across the network. Policy Management consists of the building blocks
shown in Figure 1.
Figure 1:
Policy Management Building Blocks.
As shown in Figure 2, during the last year enterprise IT organizations have
become notably more interested in deploying Policy Management. In particular,
Figure 2 shows the results of a survey that measured the interest in
implementing Policy Management. This survey was administered in both 1998 and
1999. Note that in 1998, roughly half of the respondents were neutral about
implementing Policy Management. However, in 1999, less than 30 percent of the
respondents were neutral about implementing Policy Management. More importantly,
the general trend was to shift away from being neutral to being interested in
implementing Policy Management.
Figure 2: Policy
Management Survey.
However, exactly how Policy Management should be implemented is a topic that is
debated in many forums, including a May 1999 article in Data Communications
magazine entitled "The Policy Makers." As mentioned in the article,
all policy-based schemes are comprised of three core components: a management
console, one or more policy servers, and network hardware. The purpose of the
management console is to set Quality of Service (QoS) policies, typically by
assigning traffic to predefined service classes.
The policy server is responsible for telling the switches and routers how
to handle different types of traffic in order to meet the specified QoS levels.
The switches and routers use internal queuing mechanisms to enforce the policies
and to mark the traffic so that other pieces of network equipment can do the
same.
The Data Communications article also details the issues that enterprise IT
organizations need to be aware of when evaluating a policy management solution.
Those issues include:
Products vs. slideware
Many vendors have yet to ship any products that actually enable policy
management.
Location of the Policy Server
In some implementations the policy server is embedded in the network equipment.
In others, it is a stand-alone device.
Policy Server Architecture
Some vendors deploy a two-tier architecture in which a server manager is used to
download policies to a policy server. Also, most vendors allow for the
deployment of redundant policy servers.
4.2 Service Level Management (SLM)
As mentioned, the key components of managing a converged infrastructure are
policy management and service level management. There are two reasons that would
drive an IT organization to implement service level management. The first reason
is to better link the goals of the IT organization with those of the business by
clearly defining the services that the IT organization provides to the Business
Units. The second reason is to provide a feedback mechanism for policy
management.
To link the goals of the IT organization with those of the business, there needs
to be various levels of Service Level Agreements (SLAs), both business and
technology focused, and these need to relate to each other. For example, one set
of services that an IT organization may offer to the enterprise is conferencing
services. This could potentially include audio-conferencing, desktop
video-conferencing, and room video-conferencing.
The service description for this set of services might include
user-centric metrics such as availability and the cost of the service. However,
these user-centric metrics need to be mapped to the service level provided by
the underlying infrastructure. In addition, once the SLAs have been established,
it is incumbent on the IT organization to deliver consistent and reliable
network and application performance by monitoring the environment for its
ability to meet the SLAs.
While there has been a great increase in terms of the number of IT organizations
that offer formal SLAs to the Business Units, the percentage of such IT
organizations is still less than half of the total.
However, few IT organizations work in an environment where there is no
expectation on the level of service that the IT organization offers to the
Business Units. Hence, virtually all IT organizations offer SLAs. The only
question is whether the SLA is implicit or explicit and which is best for the IT
organization and the business.
The principal reason that Service Level Management fails to be as impactful as
it might be is that IT organizations do not deploy a structured approach. The
author would like to suggest that the following are key components of a
successful deployment of Service Level Management:
_ Service
Screening and Planning
_ Service
Design and Testing
_ Service
Introduction
_ Ongoing
Service Monitoring
_ Service
Retirement
Some of the tasks listed above, such as service planning, requires a level of
skill which is often missing from an IT organization. Hence, if an IT
organization wants to be successful with service management, it needs to evolve
the skill mix in the organization.
In addition to evolving the skills of the organization, new tools and agents are
required. Nortel Networks recognizes this and it is the company's intention to
improve the effectiveness of Service Level Management (SLM) via the use of
Optivity SLM. Optivity SLM provides proactive planning, comprehensive trend
analysis, and the establishment of performance baselines to strategically and
effectively plan network growth and future enhancements. Optivity SLM enables IT
professionals to gather and aggregate application performance and availability
data directly from users. It does this via a variety of agents that are located
at both local and remote desktops. Having agents at the desktops means that
Optivity SLM can quantify the end-user experience for each network and
application transaction. Optivity SLM currently supports more than 25
prepackaged applications including popular groupware, e-mail, Web, database, and
business applications such as SAP R/3 and PeopleSoft. It is easy to monitor any
custom application by entering the name of the application and the relevant TCP
socket. It is also possible to enter a range of ports necessary for certain
categories of applications, such as Enterprise Resource Planning (ERP)
applications.
5.0 The Converged WAN
There are a variety of ways that an enterprise can benefit from a converged WAN
infrastructure. For example, many companies can save money by deploying
converged WAN access. By converged WAN access is meant the dynamic combination
of voice, video, and data traffic onto a single transmission group from an
enterprise location to the carrier's Point of Presence (POP). To understand how
converged WAN access can help to reduce costs, consider a company that has 200
branch offices with multiple T1s from each branch office to the carrier's POP.
Assume that the average monthly cost of these T1s is $500 and they are
statically configured to carry voice, SNA, and IP traffic. Further assume that
by deploying Inverse Multiplexing over ATM (IMA) the enterprise reduces one T1
per office. This results in a $100K monthly savings, or $3.6 million over a
three-year life cycle. If within that company, every dollar of revenue results
in 10 cents flowing to the bottom line, then this $3.6 million savings has the
same impact on the company's bottom line as $36 million in new revenue.
A converged WAN backbone can help an enterprise to both save money and better
compete in the marketplace. Nortel Networks is a good example of a company that
has achieved these two goals by deploying an ATM wide area network to carry
voice, data, and video traffic. Five years ago, Nortel Networks had seven
separate but overlapping networks serving the voice, data, fax, and video needs
of roughly 60,000 users in both Nortel Networks and Bell Northern Research
locations worldwide. At the core of these networks were a multitude of TDM
devices supporting relatively low-speed lines; i.e., a Fractional T1 up to a
couple of T1s. The predominant user of bandwidth at the time was voice traffic.
Two of the goals that Nortel Networks had when they deployed their converged WAN
backbone were:
_ To provide
increased bandwidth to meet the explosive growth in network traffic, while
facilitating the deployment of emerging bandwidth-intensive business
applications
_ To consolidate
the corporation's existing enterprise networks into a single WAN as a means of
trimming costs and simplifying network management
The network that Nortel Networks deployed in the U.S. was based on
carrier-provided ATM services running at DS-3 speeds. The network also had
private T3s between the U.S. and Canada, as well as between Ottawa and Toronto.
The original business case for this network focused on combining multiple data
networks onto a common WAN. However, the business case that was accepted by
Nortel Networks senior management called for combining Nortel Networks' voice
traffic with its data traffic. By combining the voice and data traffic onto a
common WAN, the payback period for this network was just over a year. Part of
the reason that Nortel Networks was able to achieve such a quick payback was
that it combined standard voice compression techniques with speech activity
detection, which eliminates the pauses in conversations. The result was a
significant reduction in the transmission capacity required for the voice
traffic.
During the last five years, Nortel Networks' converged WAN has facilitated both
great growth and changing business models. For example, in that timeframe, the
voice traffic on the Nortel Networks backbone has nearly doubled. However, at
the same time, the data traffic on the backbone has doubled three times! While
it most likely would have been possible to support this traffic using TDM and
private lines, it would have been much more expensive. In addition, as recently
as three or four years ago, it was still taking Nortel Networks sixty to ninety
days to get an additional T1 installed. That would have made it difficult if not
impossible for Nortel Networks to support rapidly escalating traffic demands in
a timely manner with the old network structure. However, with their ATM
backbone, adding bandwidth can be done quickly in software.
In part because of the lower cost structure of this WAN backbone, Nortel
Networks has been able to deploy new applications that facilitate new ways of
doing business. For example, Nortel Networks currently has roughly 1.5 million
Web pages spread over 1,300 Web servers. Now when Nortel Networks announces a
product or a product enhancement, the sales force can pull down the appropriate
Web page immediately. In part because of these successes, the network now has
been expanded. There are now links between North America and the UK as well as
between the UK and the continent of Europe. In addition, there is an ATM WAN
within the UK as well as across Europe.
6.0 Summary and Next Steps
As we enter the new millennium, the competitive pressures on businesses will
only increase. In addition, over the next several years, the success or failure
of a business will increasingly be linked to the success or failure of the IT
organization. Given these pressures, it is critical that the IT organization
expand their efforts to align the goals of the IT organization with those of the
business. Sections 2 and 3 of this paper gave some insight as to how to make
that alignment. In particular, the IT organization needs to be able to
articulate the primary issues driving the business. Then for each issue, it must
then identify how the use of IT can alleviate that issue.
What is so appealing about the deployment of a converged infrastructure is that
the benefits it offers match up so well with the most common issues that
currently impact business.
Organizations looking to deploy a converged infrastructure would be well advised
to:
_ Crisply
identify what business situation or situations would benefit from a converged
infrastructure. Note that the phrase "business situation" could refer
to an issue faced by the enterprise; i.e., establishing better linkage with
customers. Or, it could refer to the business of IT, i.e., improve configuration
management by deploying Policy Management.
_ Define which
component or components of convergence is the most appropriate. In particular,
analyze if converged WAN access can be beneficial.
_ Provide a
solution that will meet the expectations not just of performance but also of
functionality, if IP Telephony is intended to be part of the convergence
solution.
_ Deploy the
infrastructure now that can support Policy Management and Service Management,
even if deployment of this functionality is not intended in the near term.
_ Define the
services that IT offers to the enterprise. For each service, list a few key
performance and economic parameters.
_ Implement
the structure to begin to market the services provided by IT to the enterprise.
_ Develop and
implement a plan to ensure that the skills exist within the organization to
interface with the Business Units on a more strategic basis as well as to
implement and manage the new technologies.
*Nortel Networks, the Nortel Networks logo, the Globemark, How the World Shares
Ideas, Unified Networks, and Accelar are trademarks of Nortel Networks. All
other trademarks are the property of their owners. (c) 2000 Nortel Networks. All
rights reserved. Information in this document is subject to change without
notice. Nortel Networks assumes no responsibility for any errors that may appear
in this document. Printed in USA.