Ethernet: The Definitive Guide (Definitive Guides) [Paperback]

Ethernet--the predominant technology binding together networks over the last score of years--is well understood but seldom clearly explained. Ethernet was conceived as a way to arbitrate access to a shared communications channel, facilitating not only broadcast and reception, but actual inter-and infra-operating system communication between physically separate computers. The end product of the Aloha Network, a practical experiment in Hawaii in the 1970s, Ethernet took root early in the 1980s and thrives today in much more complicated, evolved form. The author brings a wealth of practical experience, stemming back to Ethernet's creation, in developing, designing, installing and supervising Ethernet-based networks of incredibly varied size, scope and power.

Ethernet: The Definitive Guide is divided into five sections, providing a full-scale, vivid introduction to Ethernet, and featuring its fascinating history as well as an easy-to-understand survey of its complex year by year and accumulating a great deal of inbred language that can baffle the newcomer. This book is ideal as a resource for such a person, as well as being a handy one-stop reference for the more seasoned professional. The author has an easy style that puts the complexities into context, and his sage advice reassures while building knowledge about Ethernet (and related standards). The index is very helpful, and turns the book into a useful reference manual long after it has been used to provide an overview of the subject.--Wilf Hey

'The book is written in a straightforward manner and along the way it debunks some of the myths surrounding Ethernet, such as the notion that collisions are bad. If you have any involvement with Ethernet, whatever the scale, this book will be an invaluable resource that will allow you to tackle any future Ethernet projects with confidence.' - Pete Martin CEng, MBCS, Chief Programmer, Marconi Communications - Computer Bulletin, March 2001 (5 stars)

This super book shows how Ethernet components can be combined to create Ethernet LANs.

Charles E. Spurgeon is the senior network architect at the University of Texas at Austin, overseeing a network that supports more than 30,000 computers. He has worked on large campus networks for 20 years, beginning in 1979 when he used to "test" network equipment by using it to read the Science Fiction Lover's Digest over the ARPANET. In the 1980s he helped develop the Stanford University network, which included building a number of prototype Cisco Systems routers and linking them together with Xerox PARC's experimental Ethernet running at 3 Mbps. In 1994, he created what is widely acknowledged as the most comprehensive and popular Ethernet Web site (http://www.bellereti.com/ethernet/ethernet.html). Charles, who attended Wesleyan University, lives in Austin, Texas with his wife, Joann Zimmerman, and their cat Sophie. In his spare time he reads seafaring novels, science fiction, mysteries, and anything else he can find on the shelves of their 4,000 volume home library. This is his third book on Ethernet.

From Chapter 13: Multi-Segment Configuration Guidelines

The individual media chapters you've just read covered the basic configuration guidelines for a single segment media system. However, when it comes to building a more complex half-duplex Ethernet system based on repeater hubs, you need to know what the multi-segment guidelines have to say.

The official configuration guidelines provide two approaches for verifying the configuration of a half-duplex shared Ethernet channel: Transmission System Model 1 and Transmission System Model 2. Model 1 provides a set of "canned" configuration rules. As long as your half-duplex network system meets these basic rules, it will function correctly in terms of the essential timing specifications. Model 2 provides a set of calculation aids that make it possible for you to evaluate more complex network topologies that aren't covered under the Model 1 configuration rules.

This chapter describes the rules for combining multiple segments with repeater hubs to build complex half-duplex Ethernet systems operating at 10-, 100- and 1000-Mbps. We begin by looking at the scope of the configuration guidelines. To help make it clear how the guidelines apply to a single Ethernet system, we then need to describe the function of a collision domain. Following that, we describe the Model 1 and Model 2 rules as they apply to each Ethernet system.

Scope of the Configuration Guidelines

The configuration guidelines apply to the Ethernet equipment described in the IEEE 802.3 standard. Further, the Ethernet media segments must be built according to the recommendations in each media system standard. If your half-duplex network system includes Ethernet equipment or media segments that are not described in the standard, you may not be able to use the configuration guidelines to verify its operation.

The engineers in the IEEE committee developed the configuration rules based on the known signal timing and electrical performance specifications of Ethernet equipment that fully conforms to the published standard. That way, they could predict what the behavior of the Ethernet equipment would be, and how the signal timing would function across multiple segments.

Using non-compliant equipment and media segments makes it impossible to evaluate the timing of the network. Linking media segments together with equipment not described in the standard also makes it impossible to evaluate the timing. In both cases, there is no way for the design engineers to know how such equipment and media segments will behave. While such an Ethernet may function perfectly well, it will be "outside the standard," and you will not be able to use the official IEEE configuration guidelines to verify that such a system meets the half-duplex timing specifications.

Network Documentation

When it comes to evaluating the configuration of your network and for later troubleshooting, you should document each network link in your system when it is installed. The documentation should include the length of each cable segment in the link, including any transceiver cables and patch cables. Also included should be the cable type used in each segment and any information you can collect on the cable manufacturer, the cable ID numbers printed on the outer sheath, and the cable delay in bit times provided by the manufacturer. The standard recommends that you create your own forms based on Table 13-1 for use in collecting information and documenting your network.

[Table 13-1: Sample Cable Segment Documentation Form]

Collision Domain

The multi-segment configuration guidelines apply to a half-duplex Ethernet collision domain described in Chapter 3, The Media Access Control Protocol. A collision domain is formally defined as a single Carrier Sense Multiple Access with Collision Detection (CSMA/CD) network in which there will be a collision if two computers attached to the system transmit at the same time.

An Ethernet system composed of a single segment or of multiple segments linked with repeater hubs constitutes a single collision domain. Figure 13-1 shows two repeater hubs connecting three computers. Since only repeater connections are used between the segments in this network, all of the segments and computers are in the same collision domain.

[Figure 13-1. Repeater hubs create a single collision domain]

Another important point is that all segments within a given collision domain must operate at the same speed. That's because repeater hubs assume that all segments connected to the repeater are operating at the same speed and have the same round-trip timing constraints. This is also why there are three sets of half-duplex configuration guidelines, one each for 10-, 100-, and 1000 Mbps Ethernet. Each of the three Ethernet speeds has its own round-trip timing constraints and its own set of configuration guidelines.

The configuration guidelines described in this chapter are taken directly from the IEEE 802.3 standard, which describes the standards for the operation of a single half-duplex Ethernet local area network (LAN). Therefore, these guidelines only apply to a single collision domain, and have nothing to say about combining multiple Ethernet collision domains with packet switching devices such as switching hubs or routers. Switching hubs enable you to create new collision domains on each port, allowing you to link many networks together. You can also link segments operating at different speeds with switching hubs. The operation and configuration of switching hubs are described in Chapter 18, Ethernet Switching Hubs.

Model 1 Configuration Guidelines for 10 Mbps

The first configuration model provided in the 802.3 standard describes a set of multi-segment configuration rules for combining various 10 Mbps Ethernet segments. Bold text is taken directly from the IEEE standard.

* Repeater sets are required for all segment interconnection. A "repeater set" is a repeater and its associated transceivers (i.e., medium attachment units, or MAUs) and attachment unit interface (AUI) cables, if any. Repeaters must comply with all IEEE specifications in clause 9 of the 802.3 standard, and are used for signal retiming and reshaping, preamble regeneration, etc.

* MAUs that are part of repeater sets count toward the maximum number of MAUs on a segment. Twisted-pair, fiber optic and thin coax repeater hubs typically use internal MAUs located inside each port of the repeater. Thick Ethernet repeaters use an outboard MAU to connect to the thick coax.

* The transmission path permitted between any two DTEs may consist of up to five segments, four repeater sets (including optional AUIs), two MAUs, and two AUIs. The repeater sets are assumed to have their own MAUs, which are not counted in this rule.

* AUI cables for 10BASE-FP and 10BASE-FL shall not exceed 25 m. (Since two MAUs per segment are required, 25 m per MAU results in a total AUI cable length of 50 m per segment.)

* When a transmission path consists of four repeaters and five segments, up to three of the segments may be mixing and the remainder must be link segments. When five segments are present, each fiber optic link segment (FOIRL, 10BASE-FB, or 10BASE-FL) shall not exceed 500 m, and each 10BASE-FP segment shall not exceed 300 m. A mixing segment is defined in the standard as one that may have more than two medium dependent interfaces attached to it (e.g., a coaxial cable segment). A link segment is defined as a point-to-point full-duplex medium that connects two and only two MAUs.[3]

* When a transmission path consists of three repeater sets and four segments, the following restrictions apply:

* The maximum allowable length of any inter-repeater fiber segment shall not exceed 1000 m for FOIRL, 10BASE-FB, and 10BASE-FL segments and shall not exceed 700 m for 10BASE-FP segments.

* The maximum allowable length of any repeater to DTE fiber segment shall not exceed 400 m for 10BASE-FL segments and shall not exceed 300 m for 10BASE-FP segments and 400 m for segments terminated in a 10BASE-FL MAU.

* There is no restriction on the number of mixing segments in this case. In other words, when using three repeater sets and four segments, all segments may be mixing segments if desired.

Figure 13-2 shows an example of one possible maximum Ethernet configuration that meets the canned configuration rules. The maximum packet transmission path in this system is between station 1 and station 2, since there are four repeaters and five media segments in that particular path. Two of the segments in the path are mixing segments, and the other three are link segments.

[Figure 13-2. A maximum Model 1 10 Mbps configuration]

While the canned configuration rules are based on conservative timing calculations, you shouldn't let that lead you to believe that you can bend these rules and get away with it. Despite the allowances made in the standards for manufacturing tolerances and equipment variances, there isn't a lot of engineering margin left in maximum-sized Ethernets. If you want maximum performance and reliability, then you need to stick to the published guidelines.

In addition, while the configuration guidelines emphasize the maximum limits of the system, you should beware of stretching things as far as they can go. Ethernets, like many other systems, work best when they aren't being pushed to their limits.

The "5-4-3" Rule

An over-simplified version of the 10 Mbps Model 1 rules, called the "5-4-3" rule, has been circulating for some years. Various forms of the 5-4-3 rule have been published, and some of them include misleading terms that are incorrect. To quote from one widely distributed configuration guide, the 5-4-3 rule means that there may be as many as five segments connected in series in a network. This guide further states that up to four repeaters may be used, and up to three "populated segments." A populated segment is defined as a segment that is "attached to PCs."

While this may sound like an easy to remember rule of thumb, the "5-4-3" rule is an over-simplification of the actual configuration rules described above. Worse, the use of the term "populated segment" is misleading. This definition means that a coax segment could be regarded as an "unpopulated" segment in a network system as long as two conditions were met. First, the coax segment was not used to support PCs and, second, the segment was only used as a link segment to connect to a repeater at each end. However, this is incorrect.

A link segment is specifically defined in the 802.3 standard as a segment based on a point-to-point full-duplex media type that connects two-and only two--MAUs. A full-duplex medium means that the medium provides separate transmit and receive data paths. This is important, since collision detection occurs faster on a full-duplex medium than it does on coaxial segments. This difference in timing is factored into the total round-trip timing delays that are incorporated in the Model 1 configuration guidelines. That's why the notion of an "unpopulated" coax segment that could be used as a link segment is misleading and incorrect.

To recast the 5-4-3 rule into something closer to reality, we can define it to mean that you can have up to five segments in series, with up to four repeaters, and no more than three "mixing" segments. If three mixing segments are used, then the remaining two segments must be link segments as defined in 802.3. Actually, you can have up to four mixing segments under some circumstances as described in the real 802.3 rules above, so even our corrected 5-4-3 rule is still an over-simplification.