Hagar, ( PART 1)The Networking dilemma:I think you make a good point and perhaps [ definitely :-) ]I've oversimplified the point.
We are dealing with complex issues of everchanging new technology and needs and how to address it. I think this is why the ATT/IBM/NN deal looks so exciting: you have the converging expertise of people that have experience in managing large networks to get you the best deal. It would be nice if all this could be pretested in toto in some testing center.
This is why I like theATT/IBM/NN deal
This older article ( 1995 ) foresaw all this:
"Once the cost of the network is well understood, then one key technique that many network managers use to deal with this situation effectively is the strategic use of complementaryout-tasking. The manager must identify where out-tasking could add value to the organization. This requires, first, an analysis and understanding of current costs, of where people within the organization can add the most value, and of which functions, when performed in-house, either do not add value or are without strategic impact. Second, the manager needs to identify areas where the skillful use of outside partners can assist in making strategic decisions more effectively. An out-tasking strategy can then be pursued within the framework of corporate budgets and goals.
TA
------------------------------------------------------------
nwfusion.com
PART 1
The Networking Dilemma:
How to Identify and Successfully Manage the Costs and Complexities of Communications
Jane Bradshaw
Digital Equipment Corporation
James Metzler
Strategic Networks Consulting
THE IMPLICATIONS OF UNPRECEDENTED GROWTH:
EXECUTIVE SUMMARY
These are the most challenging times in networking history, and, as every enterprise network manager is aware, effective leadership requires working smarter, rather than simply working
harder. Coping with today's complex networking issues can be overwhelming. Managers are dealing with unprecedented growth in the number of users they support, increased bandwidth
requirements due to traffic-intensive applications, and an ever-increasing range of technologies and vendors from which to choose. Meanwhile, budgets remain flat or include only modest
increases and finding and retaining qualified staff is increasingly difficult.
In respect to managing the LAN and WAN effectively, unprecedented growth - coupled with strict budget constraints - means that continual evaluation of different technologies and services
is a must. This includes understanding the technology and how it fits in the environment, selecting vendors, evaluating prices, designing and implementing the network, training and deploying
employees, and managing and optimizing the system.
To cope in such a dynamic environment and manage successfully within it, two concepts must be understood: the cost of network ownership and the cost of complexity. First, the major cost
components of the network (people, facilities and equipment) must be clearly delineated. Then, the costs of the network as it functions as a dynamic interaction between people, facilities, and
a wide array of technologies, equipment, and vendors, must be analyzed. This is the cost of complexity. After these costs have been translated into bottom-line impact, the manager must
maximize the use of each component and minimize the costs of networking through the skillful orchestration of internal resources and complementary out-tasking.
FACTORS DRIVING UNPRECEDENTED GROWTH
The industry's unprecedented growth is driven by two key factors. The first factor is advances in technology that impact the way business is accomplished; for example, client/server
computing, data warehousing, groupware, document and image management, multimedia applications, mobile and wireless communications, teleconferencing, and others. One of the
challenges facing network managers is that in many cases the people who are driving business change often assume that the network to support these changes is either already in place or can
be deployed at little to no cost.
The second factor is the accelerated rate of change in the way corporations do business - new approaches such as corporate re-engineering, empowered management, self-directed teams,
and the extension of the network to include key customers and suppliers. The changes brought on by this second factor - a tremendous increase in the number of users and applications,
support of a growing range of computing styles, demand for different technologies in both local and wide area networks, shortages of good people, budget constraints - put greater demands on
the components of the first factor, technological advance. The result of these two factors is a significant increase in the level of demand on management skill.
THE PLIGHT OF THE NETWORK MANAGER
The demands on network managers continue to escalate. Network managers are faced with the need to support a growing user population, an increased range of computing styles, numerous
alternative technologies, shortages of trained staff, and stringent budget limitations. Network managers must ensure that the network provides the following:
Continuous up time
Improved response time
Reduced deployment time
Improved price/performance
In addition to meeting the demands of the daily operation of the existing network, managers must constantly address the issue of evolving the network to support emerging applications. For
example, many network managers must support existing legacy Systems Network Architecture (SNA) applications while new styles of computing, such as client/server, image and
multimedia-based applications, are proliferating. Typically, the newer applications are consuming greater quantities of bandwidth and are demanding different Quality of Service (QoS)
guarantees. This forces the network managers to address the issue of integrating existing network technologies with emerging and future network technologies which have the promise of
providing the required bandwidth and QoS capabilities.
To make matters worse, there are a vast array of alternative technologies. Some of these include:
Traditional shared media such as Ethernet and Token Ring
High-speed shared media such as Fast Ethernet, FDDI/CDDI, 100VG-AnyLAN
LAN switches for high-speed bandwidth to users on a per port basis
Traditional leased lines ranging from DS0 (56/64Kbps) to T3 (45Mbps)
Integrated Services Digital Network (ISDN)
Frame relay
Switched Multimegabit Data Service (SMDS)
Asynchronous Transfer Mode (ATM)
BUDGETING AS A MANAGEMENT TOOL
Budget constraints play a key role in the decision-making process for developing and upgrading the network. For example, a survey of network managers shows that their budget goals for
1996 were either to control growth or keep the budget flat while increasing bandwidth (see Figure 1). Typically, the network manager is asked to accommodate increases in data traffic of
30% to 50% while working with budget increases that average 3%. In addition, new technologies are a double-edged sword. While new technology typically offers the potential to improve a
network's price/performance, the evaluation and eventual deployment of new technology consumes increasingly scarce resources.
Figure 1. Network Managers Budget Goals
A PLANNING STRUCTURE
The CIO of a large medical organization recently expressed his current situation as where he "gets to be the one to say no we can not do that and incidentally, I need more money in my
budget". In order to avoid these no-win situations, network managers should institute the following continuous planning cycle:
1.Quantify the current status of the network
2.Identify the specific forces that are impacting the network
3.Determine the set of viable network alternatives
4.Identify the decision criteria
5.Evaluate the alternatives based on the criteria
6.Develop a business case for the network upgrades
TWO KEY CONCEPTS FOR EFFECTIVE MANAGEMENT
A key component of each of the six steps is cost. In particular, network managers need to understand what is driving their bottom line costs. Once these drivers are understood, decisions can
be made on where to invest time and spend limited budget dollars. Network managers need to understand two important concepts: the cost of network ownership and the cost of network
complexity. Any investment, of either time or money, should be focused on these main cost drivers of the network. Once these concepts are understood, network managers will have a better
idea of where the majority of their costs occur and can focus on providing services that add the most value to their organizations.
To better understand and manage the costs involved, the remainder of this paper concentrates on the costs of network ownership and network complexity and make suggestions on how
network managers can deal with them effectively.
THE COST OF NETWORK OWNERSHIP
The cost of network ownership identifies the cost drivers in a network. In the local area network (LAN), about 47% is spent on people costs and 39% on capital equipment. In the wide area
network (WAN), people costs are about 42% while facilities charges (mainly circuit costs) constitute 33%.
In terms of people costs, a booming industry means that networking professionals are commanding and getting higher salaries. Increased demand has made finding experienced technical staff
a challenge. Any classified section in the newspaper, web site, or recruitment firm, makes it obvious that people with networking skills are in high demand and very difficult to find. After they
are hired, network managers have the added challenge of retaining them while keeping them current on upcoming changes in the industry.
AN ANALYSIS OF THE COSTS OF NETWORK OWNERSHIP
Although many network managers have a general understanding of the primary cost components of their networks, the majority do not know the associated costs in detail. The three major
components of cost involved in a network include:
Capital equipment - primarily hardware and software
Technical support staff - salaries and fringe benefits
Facilities - leased lines, switched circuits, electrical power and floor space
In the local area network, the facilities costs (power, floor space, but not circuit charges) tend to be minimal compared to people and hardware costs. (See Figure 2). Network managers who
focus on increasing staff productivity are more likely to meet their business goals. Since capital equipment in the LAN represents a significant portion of the total cost, care should be taken
when evaluating new equipment to be installed on the LAN. The successful strategy is to invest more in capital equipment if it allows the network manager to leverage personnel. At a
minimum, the equipment should have the ability to be remotely managed via SNMP. Also, a manager evaluating LAN switches should make RMON and the upcoming RMON II a major
purchasing criteria. The payback in network management efficiencies could be measured in just a few months.
Figure 2. Local Area Network: People, Facilities and Capital Costs
A useful metric for network managers is the annual LAN cost per desktop, which can range between $2,800 and $3,500. One factor that impacts where in this range a particular company fits
is the type of technology used within that company's LAN infrastructure. For example, the hardware cost for a shared 10 Mbps Ethernet LAN will cost less than the hardware cost for a
dedicated 10 Mbps Ethernet LAN. A second factor that will impact the annual LAN cost per desktop is the level of support provided. In particular, companies that provide a high level of
support tend to require more support staff.
The costs for supporting a WAN tend to be notably different than the costs for supporting a LAN. For example, facilities costs in the WAN, which are heavily influenced by circuit charges,
are a monthly recurring cost. Figure 3 illustrates the breakdown of costs in the WAN. One interesting conclusion that can be drawn is that people and facilities costs associated with WAN
operations constitute 75% of the total. Although it is not reasonable to conclude that equipment is unimportant; in order for network managers to control their costs in the WAN, they should
focus their energies on reducing facilities costs and increasing staff productivity.
Figure 3. Wide Area Networking: People, Facilities and Capital Costs
Managers of international networks should pay very close attention to their wide area transmission costs. These costs tend to increase dramatically when companies extend their networks
outside the U.S. For example, the discounted price a mid-to-large sized company would pay for a 1,000 mile 56 Kbps circuit in the U.S. is approximately $750 per month. Unlike domestic
private lines, international private lines are typically priced as two half circuits. Table 1 contains the monthly cost that a mid-to-large sized company could expect to pay for selected 64 Kbps
international circuits. One conclusion that can be drawn from Table 1 is that the cost of an international private line can be an order of magnitude more than the cost of a domestic private line.
Originating City
Half Circuit Cost
Terminating City
Half Circuit Cost
Total Monthly Cost
Amsterdam
$4,100
Geneva
$2,800
$6,900
Bangkok
$6,000
Chicago
$4,000
$10,000
Bogota
$5,500
New York
$2,800
$8,300
Hong Kong
$5,500
Chicago
$4,000
$9,500
Table 1. Monthly Cost for Selected International Private Lines
(December 1995):
Another useful metric for network managers is the cost to move a megabyte of data across their WAN. In particular, network manages could expect to pay between $0.50 and $75 to
transfer a megabyte of data across a corporate WAN. This broad range in the cost of transmitting a megabyte over the WAN is driven by two factors. The first of the factors is described
above; i.e., the wide range in the cost of WAN links based on whether they are local, national, or international circuits. The second factor is the wide variation in the utilization levels of WAN
circuits.
THE COST OF NETWORK COMPLEXITY
A primary way to lower the cost of support is to reduce complexity in the network. Effective LAN and WAN management requires continual evaluation of a tremendous number of
technologies and services. This includes everything from understanding the advantages and disadvantages of each technology or service, selecting appropriate vendors, evaluating pricing of
the equipment or service, performing network design, planning the implementation, obtaining funding, deploying the technologies, and training the operations group and end users. Then it is
time to manage and optimize the service provided. How can the manager succeed in this environment? He or she must identify the core competencies of the organization, place people in the
organization where they can add the most value, and determine if there is another way to provide the service at a lower cost. Once this is done, it will become clear what activities should be
provided in-house versus what should be out-tasked.
AN ANALYSIS OF THE COSTS OF NETWORK COMPLEXITY
After the cost of network ownership is calculated, the cost of network complexity must be factored into the budget. What are the causes of network complexity? The main complexity driver
is the implementation of different alternatives for each of the components of the network. The components of the network include:
WAN services
Vendors
Protocols
LAN technologies
Network Operating Systems
Network Management Systems
The major cost associated with complexity is the burden that it places on support staff, as opposed to the related increase in capital equipment and facilities costs. This is particularly
troublesome since people costs are one of the biggest budget items associated with running a network (as shown in Figures 2&3). Included are the "visible" costs such as salaries and benefits
or typical budget sheet items. Also included are the "invisible" costs that do not show up in a budget, such as "resident guru" costs (the one person in the group who "knows it all" and spends a
lot of time troubleshooting problems rather than attending to assigned responsibilities). The result of this complexity is that support staff efficiency decreases, and it sets in motion a series of
events that ripple out of the network group into nearly all IT departments and business units.
Consider the example of remote office networking. Previously, traditional host-based networks required support from the data center. Users at remote sites required no software and they
needed to access only a few applications. This led to a fairly static wide area network. Today the remote office LAN is continually changing as new devices, users and applications are being
added while WANs become unpredictable regarding traffic patterns and bandwidth requirements. Layered on top of this are various network components that are supplied by multiple
vendors, each utilizing versions of various "standards," each requiring their own management application, software support and user training. Increased complexity can lead to network
gridlock, when managers are afraid to make a change for fear of possible side effects. If not managed effectively, the complexity issues engendered by using numerous vendors can be
devastating to business operations.
When network managers upgrade their LANs in response to user demand, it invariably leads to more traffic over the WAN. The good news is that as network managers plan their WAN
upgrades they have a variety of technologies and services from which to choose. However, as shown in Table 2, each of these services has advantages and disadvantages. There is no one
service that is best for all applications. In order to minimize costs, network managers need to evaluate which transmission services provide the best support for their suite of applications.
Service Class/ Services
Advantages
Disadvantages
Circuit Switched
Voice Services:
POTS (Plain Old Telephone Service), SDN
(Software Defined Networks),
800 Services
Widely Deployed
Highly Interoperable (POTS)
Superior Network Management (SDN)
Limited Throughput
Slow Call Setup
Non-capped Pricing
Circuit Switched
Data Services:
ISDN (Integrated Services Digital Network),
Switched 56
Appropriate for Video & Other High-speed Intermittent
Usage Applications
Quick Call Setup (ISDN)
Widely Varying Pricing
Varying Deployment & Ease of Installation
Non-capped Pricing
Frame Services:
frame relay, X.25
Capped Pricing
Ability to Support Bursty Traffic
Widely Deployed
Varying Tariff Structures
Limited Interoperability
Cell Services:
SMDS (switched multimegabit data services),
ATM (Asynchronous Transfer Mode)
Very High Speed
Ability to Support Bursty Traffic
Embryonic Technology
Limited Interoperability
Limited Deployment
Typically Custom-based Pricing
Dedicated Services:
Private Line
Varying Speed
Widely Available
Highly Interoperable
Potentially Wasted Bandwidth
Table 2. Characteristics of Existing and Emerging Transmission Services
Currently the two most popular services for internetworking LANs in the United States are private lines and frame relay. While frame relay has come on strong in the last two years, it is not
necessarily the most cost effective service for all applications. As is generally the case with Inter-Exchange Carriers (IXCs), the cost of a private line consists of a fixed component and a
mileage sensitive one. For example, the list price of a 100-mile, 64 Kbps circuit from AT&T is $296 per month. The list price of a 1,000 mile, 64 Kbps circuit from AT&T is $575. The pricing
of AT&T's U.S.-based frame relay service is unrelated to either distance or usage, but is based on ports and Permanent Virtual Circuits (PVCs). The cost of a 56 Kbps frame relay
connection is $546 for the two ports plus a varying amount for the PVC, depending on the speed. Assuming a 16 Kbps PVC, which adds $31, the total cost ends up being $577. The
conclusion is that if the distances between the sites being connected is greater than 1000 miles, frame relay is probably going to be a less expensive solution.
Not only do network managers have to choose between transmission services such as private lines and frame relay, they also have to choose between vendors of a given service. Vendors
often do not price their services in a consistent fashion. Regulations and tariffs are very complex. For example, as previously mentioned, AT&T's U.S.-based frame relay service is not related
to distance or usage parameters. However, MCI markets a frame relay service that is both distance and usage sensitive. Therefore, in order to minimize cost, network managers need to have
a good understanding of their traffic. They must then analyze and compare differing services, such as private line and frame relay, as well assess the same service, such as frame relay, from
different vendors.
Another variable that network managers need to evaluate is that some carriers offer what is referred to as a managed frame relay service. Recently, a large US bank wanted to compare the
economic viability of deploying a managed Frame Relay service vs. deploying a frame relay service and providing the management themselves. The bank submitted a Request for Information
(RFI) to several vendors of both managed and non-managed Frame Relay services. By comparing the responses, the bank could determine the cost of having the carriers provide the
management. However, since the bank could not determine the internal cost of providing that management, it was prevented from making a decision between the two approaches based on
hard economic data.
The issue of comparing services between vendors is further complicated by the U.S. Telecommunications Act of 1996, which will increase the number of providers of both local and long
distance services. Internationally, continued deregulation of the PTTs (the government organizations providing telecommunications services), along with the 1998 European Union deregulation
of voice services, will give birth to more transmission vendors. For example, the United Kingdom's transmission market had long been dominated by British Telecom and Mercury. Due to
deregulation, there are now a myriad of vendors in the United Kingdom.
COMPLEMENTARY OUT-TASKING AS A SOLUTION
Once network managers understand the major cost drivers in their networks, they should focus their energies on efficiently managing these pieces. However, in a time of flat headcount and
annual budget increases in the 3% range, it is becoming increasingly clear that network managers will have to select core competencies that provide the highest value to an organization. They
must then identify partners to assist in those areas that are not strategic, or where in-house resources either cannot meet or cannot service the need cost-effectively. Two of the major
reasons that managers are out-tasking is to gain expertise as well as to supplement their labor to help them meet deadlines.
Network managers provide a range of interpretations of the meaning and scope of the term "out-tasking." To some, out-tasking means Virtual Private Networks(VPNs) for voice. This
practice became an attractive alternative to expensive and increasingly inflexible private Time Division Multiplexor (TDM) backbones for voice traffic in the mid 1980s. While some still view
control of voice traffic on a TDM backbone as strategically important, voice expenditures have consistently declined to less than 50% of total telecommunications budgets leading to the
increased use of VPNs for corporate voice networks.
When devising an out-tasking strategy, managers have a range of options. Out-tasking deals can go as far as the third party acquiring the corporate networking assets as well as its
network-related employees. However, that is not what we are talking about in this discussion. What we are talking about is selective out-tasking, which means that components of the network
are selected for out-tasking as part of a management strategy to accomplish business and budgetary goals. There are basically two scenarios for this type of complementary out-tasking. One involves deals where a third party will project manage such tasks as technical evaluation, installation, or roll-out of a new application such as Lotus Notes. The other involves longer-termarrangements where a third party provides services such as ongoing support and management of the network infrastructure or some piece of it. |
