Enhancing LAN Performance

Gilbert Held

• Publisher: CRC Press
• Year: 2004
• ISBN:
  9781135494285
• Edition: 1
• Topic: Computer Science
• Subtopic: Information Technology
• Index: Computer Science

Enhancing LAN Performance, Fourth Edition explains how to connect geographically separated LANs with appropriate bandwidth, the issues to consider when weighing the use of multiport or dualport devices, how to estimate traffic for new networks, the effects of configuration changes on the performance of Ethernet and Token Ring networks, the design o

Chapter 1: LAN Performance Issues

This introductory chapter focuses attention on a core set of local area network (LAN) performance issues. In doing so, it discusses why such issues are important for inter- and intra-network communications in the form of questions that we will want to answer when planning or expanding a network. Although we leave it to future chapters in this book to answer questions raised in this chapter, our discussion of performance issues serves a dual purpose. First, it makes us aware of the many factors that affect the performance of individual networking devices, as well as the networks to which they are connected. Second, it makes us aware of the fact that performance can involve the trade-off of many operating characteristics. This means that in some instances the modification of the operating characteristics of a networking device to enhance its performance may require adjustments to other features of that device and other networking devices and transmission facilities. Thus, we note that LAN performance issues in many instances involve a detailed relationship between different network components, as well as the transmission facilities used to provide a connection between separate networks.

1.1 Network Basics

Readers of this book are presumed to have a degree of familiarity with the basic operation and utilization of Ethernet, Token Ring, and ATM networks. Thus, this book was written neither as an introductory text on local area networks nor on local area networking concepts. However, because Ethernet, Token Ring, and ATM network performance requires a degree of knowledge concerning the format of the frames and their composition, Chapter 2 in this book reviews both areas.

The remainder of this chapter reviews the material in succeeding chapters with respect to the performance issues they raise. Thus, Sections 1.2 through 1.13 in this chapter can be considered previews to Chapters 2 through 13, respectively.

1.2 Ethernet, Token Ring and ATM Frame and Cell Operations

Material in Chapter 2 is presented to provide readers with a review of LAN dataflow operations. The key to understanding many performance-related issues is a detailed knowledge of the frame and cell format used on different types of LANs — material that is presented in Chapter 2.

Chapter 2 focuses attention on the fields of Ethernet and Token Ring frames, to include a discussion of Fast and Gigabit Ethernet. This will provide us with information necessary to understand the relationship between the overhead of a frame and the length of its information field. Doing so will assist us in understanding the effect of the frame length upon the efficiency of transmission on different types of LANs. Once the preceding is accomplished, we turn our attention to ATM and examine the overhead of the ATM cell and its effect on the transmission efficiency of ATM-based networks. Thus, the primary focus of Chapter 2 is to provide readers with information that can be used in later chapters to answer many LAN performance-related questions. Answers to such questions will allow us to note techniques that can be used to adjust the length of LAN frames and enhance their information transfer capability, which can result in an improvement in the efficiency of transmission over the network. Other questions we will tackle in later chapters that are related to the material presented in Chapter 2 include how to determine the effect of the frame length on bridge and router operations, as well as on their buffer memory requirements.

1.3 Estimating Network Traffic

If we already have a network in existence, there are a large number of tools we can use to determine network traffic. Once this is done, we can use the results obtained from traffic monitoring to predict the effect of a network expansion or use such information to estimate the traffic on a similar type network that is anticipated to be established. However, what can you do to estimate traffic if your organization plans to install its first local area network or a network in which the user community will significantly differ from the users of an existing corporate network? When those situations occur, you have no baseline from which you can project network traffic — a situation that can result in a significant degree of a trial-and-error unless you have a reasonable mechanism to follow to estimate network traffic.

Chapter 3 focuses on methods you can use to estimate inter- and intra-network traffic. After discussing the need for developing a realistic traffic measurement technique, this chapter focuses on the use of a traffic estimation worksheet that can facilitate the network traffic estimation process. This is followed by examples of the use of the worksheet to estimate the traffic load that will be placed on a network. In addition, this chapter examines how you can use the results obtained from the use of a traffic estimation worksheet to predict future network traffic growth, as well as how to analyze the potential effect of such growth on your existing or planned LAN. Once the traffic estimation process has been reviewed, it is used to illustrate how you can use this data to subdivide a network via the use of a local bridge or router to improve network performance. Thus, this chapter provides information concerning the traffic estimation process as well as illustrates its use to improve the performance level of a network.

1.4 Understanding and Applying Waiting Line Analysis

What is the effect on transmission between LANs when dataflow must cross a pair of remote bridges or routers? The answer to this question can be quite complex and is based on the use of waiting line analysis, which is an alternative term to the more popular expression of “queuing theory.”

Chapter 4 investigates the application of waiting line analysis to network traffic that must use remote bridges, switches, or routers to reach their destination. After reviewing some of the terms associated with queuing theory, we will apply its use to answer some key questions associated with the use of remote bridges, switches, and routers. Those questions include: What is an optimum line operating rate for those devices? What is the effect of using single or multiple port devices? And what is the effect of altering the memory capacity of a bridge, switch, or router on its servicing capacity.

Commencing in Chapter 4 we begin to use a series of programs developed using Microsoft Corporation’s QuickBASIC compiler and its Excel electronic spreadsheet program to facilitate performing a series of repetitive and tedious computations. The selection of QuickBASIC was based on the fact that it is very similar to Qbasic, which was included in MS-DOS version 5.0 and later versions of that operating system. In addition, readers with earlier versions of DOS can easily modify the QuickBASIC programs presented in this book to operate with the commonly available BASICA or GW-BASIC interpreters. Similarly, Excel represents the most popular electronic spreadsheet program currently in use. Through the creation of Excel templates, it becomes a relatively simple process for readers to modify such templates to tailor them to one or more specific operational requirements. Thus, most readers should be able to use the programs and templates presented in this book without additional cost.

To facilitate the use of the programs contained in this book, you can access them from the following Web site: http://www.crcpress.com/e_products/downloads/default.asp. At that Web address you will find three series of files you can download. One series of files has the extension .BAS and contains the source version of the QuickBASIC programs. If you have a QuickBASIC or QBasic compiler, you can easily modify those programs to reflect your specific networking environment and then compile and execute the modified program. With a little additional effort you can also use a BASICA or GW-BASIC interpreter to execute the programs presented in this book. For those readers who desire to execute programs without modification, a directly executable version of each program can be downloaded. Those programs have the same filename as the source language versions of programs presented in this book but have the extension .EXE. While you cannot modify those programs that are directly executable, many of those programs operate based upon a variable input, which allows you to adjust the use of those programs to your particular networking environment. A third series of programs you can download consist of Excel templates. The filename of each template is the same as that used in this book to facilitate their use.

1.5 Sizing Communications Equipment and Line Facilities

During the 1920s, a field of mathematics was developed to assist telephone companies in determining the number of long-distance trunks they should install between their offices to provide subscribers with a predefined capability to make long-distance calls. This field of mathematics is referred to as traffic sizing and is considered to represent a classical application of mathematics.

If we fast-forward to the present time, we find that a key area of concern of Internet service providers (ISPs), network managers, and LAN administrators is to determine how many ports or channels are required on LAN access controllers. A similar question involves determining the number of channels or ports required on a Windows-based server supporting remote access services. For both questions we can use traffic sizing to determine the number of channels or ports, modems, and access lines required to support a given subscriber base, employees that work from home, or customers who access an organization’s network. Thus, Chapter 5 focuses attention on the methodology and terminology associated with sizing communications equipment and line facilities. Through the information presented Chapter 5, you can learn how to use the scientific approach to determine the level of support required to provide dial-user access to a network in an economical manner.

1.6 Determining Availability Levels

Depending on the activities performed by your organization, you may be required to incorporate a degree of redundancy into your network structure. To do so you can simply acquire additional pairs of remote bridges or routers and connect them through the use of separate transmission facilities. As an alternative to duplicating certain network components, you can also consider the use of multiport bridges and routers, the cost of which is considerably less than separate devices. However, what is the difference in the level of availability provided by the use of dual port devices versus the use of separate devices? The answer to this question, as well as a detailed examination of the concept of availability, are the focus of Chapter 6.

Chapter 6 explains how mean time before failure (MTBF) and mean time to repair (MTTR) information — usually available from vendor product specification sheets — is used to determine both component and system availability levels. Concerning the latter, a portion of Chapter 6 illustrates how the availability level of a complex network can be reduced to a series of simple computations. Chapter 6 also includes the use of BASIC language programs and Excel templates to facilitate tedious computations and provides you with the ability to compute the availability level of devices and transmission facilities connected in series or in parallel. These programs and templates also serve as a mechanism to compute the availability level of a mixed topology network, or you can even use the information in this chapter to compare and contrast redundant and nonredundant computer hardware configurations. Using these programs, templates, or the calculation methods described in Chapter 6, you can obtain the information necessary to determine whether or not the cost of one network configuration versus another is worthwhile with respect to the additional level of network availability obtained through the use of different communications configurations.

1.7 Ethernet Network Performance

Just how fast can frames flow on an Ethernet network? While the preceding question might appear to be taken from a technically oriented game show, in actuality the answer to this question has a considerable bearing on the performance level of bridges and routers prior to those devices becoming possible network bottlenecks.

Chapter 7 focuses attention on the Carrier Sense Multiple Access with Collision Detection (CSMA/CD) network access protocol. By closely examining this protocol, we can determine the maximum frame rate that can be supported on a 10 Mbps, 100 Mbps, and even 1 Gbps Ethernet networks based upon different frame lengths. This, in turn, provides us with the ability to determine if the performance level of a bridge or router listed by a manufacturer is an appropriate decision criterion for equipment acquisition. That is, if the performance level of a device is greater than a certain frame forwarding rate, which represents the maximum frame rate that can be supported on an Ethernet network, the ability of a bridge, switch, or router to transfer frames beyond that rate is superfluous if the device only supports one communications circuit. Thus, the ability of one vendor’s bridge or router to forward frames at a faster rate than another vendor’s product may not be applicable to consider when evaluating competitive products.

Once the Ethernet frame rate is determined, Chapter 7 then presents an easy-to-use method to predict throughput between interconnected networks. This method permits you to estimate the best-case transfer time to upload or download files across connected networks, as well as to project the average time required to perform those activities.

1.8 Token Ring Network Performance

The question previously asked concerning how fast frames can flow on an Ethernet network is also applicable to Token Ring networks. That is, if you can determine the flow of information on a Token Ring network, you can use this information to estimate the performance of the network as additional stations are added to the network. You can also use this information to determine the filtering and forwarding rates required by bridges, switches, and routers connected to a Token Ring network prior to those devices potentially becoming a bottleneck and congesting the flow of data between networks. Last, but not least, by determining the frame flow on a Token Ring network, you can use this information to develop a model to project network and inter-LAN transmission time. If the inter-LAN network is to be created or was created through the use of a WAN transmission facility, you can easily adjust the model to reflect different WAN operating rates. Then, you can determine an optimum WAN operating rate that will satisfy your organization’s communications requirements without having to simply guess upon the selection of a WAN transmission facility or initiate an expensive trial-and-error process.

As might be expected from the previous paragraph, the initial focus of Chapter 8 is on the development of a model to reflect the flow of frames on a Token Ring network. Once this is accomplished, the model will be exercised to determine the frame carrying capacity of a Token Ring network under different operating conditions and network configurations.

The development of a Token Ring traffic model will require consideration of a large number of operating conditions and network configuration data. Some of the parameters that will have a bearing on the flow of frames on a Token Ring network include the number of stations on the network, the length of each lobe and the length of the ring, the average frame size, and the operating rate of the network. To facilitate our computations, we again turn to the BASIC programming language and develop several programs as well as spreadsheet models to facilitate our computations. Both the Token Ring program and spreadsheet model listings and results from executing each program are contained in this chapter. In addition, readers are referred to the use of a set of tables contained in a file at http://www.crcpress.com/e_products/downloads/default.asp that can be used to reduce the Token Ring frame flow projection process to a simple table lookup operation.

Once the Token Ring model is developed and exercised, this information is used as a foundation for determining bridge, switch, and router performance requirements. In doing so, frame flow information is used to project a range of performance that network devices should support, as well as the effect of network changes on the frame flow on a Token Ring network.

1.9 ATM network performance

Although ATM is a connection-oriented network based on the use of switches, it has certain characteristics that affect its performance. For example, what happens when too many devices connected to an ATM switch initiate communications? Chapter 9 turns our attention to ATM network performance. However, to ensure readers have an appreciation for the characteristics of an ATM network, we first review those characteristics prior to developing models and exercising those models.

1.10 Working with Images

The old adage that “one picture is worth a thousand words” could be rewritten from the network manager’s or LAN administrator’s perspective as “transporting images can ruin network...

Table of contents

• Cover Page
• Other Auerbach Publications
• Title Page
• Copyright Page
• Preface
• Acknowledgments
• Chapter 1: LAN Performance Issues
• Chapter 2: Ethernet, Token Ring, and ATM Frame and Cell Operations
• Chapter 3: Estimating Network Traffic
• Chapter 4: Understanding and Applying Waiting Line Analysis
• Chapter 5: Sizing Communications Equipment and Line Facilities
• Chapter 6: Using the Availability Level as a Decision Criterion
• Chapter 7: Estimating Ethernet Network Performance
• Chapter 8: Estimating Token Ring Network Performance
• Chapter 9: ATM Performance
• Chapter 10: Working with Images
• Chapter 11: Using Intelligent Switches
• Chapter 12: LAN Monitoring Tools
• Chapter 13: Transmission Optimization Techniques
• Appendix A: Review Problems