Routers and LANs
IP Routing Fundamentals
Author: Mark Sportack
Publisher: Cisco Press (53)
Although routers are often thought of as wide-area network (WAN) devices,
they can be equally useful in local-area networks (LANs). Routers, by virtue
of their capability to communicate at both the data link layer and the network
layer, can provide LAN administrators with a multitude of options for managing LANs and bolstering
Some of the many things that a router can do in a LAN environment include
segment a LAN's media access and MAC broadcast domains, interconnect different
LAN architectures, collapse a LAN's backbone into a single device, and provide
a gateway to the WAN. This chapter examines all these LAN-level router functions.
Before delving into the various roles that routers can play in a LAN,
it is necessary to first understand some of the functional aspects of a LAN.
This will make it easier to understand the effects that routers have when
used in a LAN.
A LAN, at its simplest, is a data communications vehicle that operates
at the physical and data link layers of the OSI reference model. LANs support
two critical functions:
The key word in the second item is local. The definition
of local may change subtly with context: A LAN has boundaries
that constrain its capability to perform these two functions. For example,
a LAN has a media access domain and a Media Access Control (MAC) broadcast
domain. Each domain forms the boundary around a group of local, LAN-attached
Broadcasts can also be performed at the network layer. These broadcasts
are bounded by the network layer domain.
A media access domain consists of all the devices connected to a
LAN that must share the LAN's bandwidth. The name and nature of this domain
depends on the media access methodology employed in a LAN. The two primary
methodologies for regulating media access are contention and token
Other media access arbitration techniques exist, but these two account
for the vast majority of existing LANs. More importantly, they will adequately
demonstrate the differences between a media access domain as well as a MAC broadcast
In a LAN that uses contention to arbitrate permission
to transmit, such as the various Ethernets, this domain is known as a contention,
domain. These names reflect the fact that this is a competition-based, chaotic,
and less-than-perfectly reliable access-arbitration technique.
In essence, any device in the contention domain may begin transmitting
if it detects no other traffic on the transmission media. The lack of traffic
is presumed to mean an idle LAN. In fact, because transmissions are not instantaneous,
silence of the LAN may just mean that a device is transmitting but that transmission
hasn't reached all the peripheral devices yet. Therefore, a device may begin
transmitting only to have its transmissions collide with another. In such
cases, both transmission streams are compromised and must be retransmitted.
In an Ethernet network, regardless of the media type or transmission
speed, a contention domain consists of all the devices that must compete for
the right to transmit. IEEE-compliant Ethernet LANs can support up to a maximum
of 1,024 station devices in a single contention domain. This means that there
can be up to 1,024 total devices competing for the right to transmit in an
Ethernet contention domain.
In LANs that pass tokens to regulate media access, such
as Fiber Distributed
Data Interface (FDDI) or Token Ring, the media access domain is called the token-passing
domain. Media access is arbitrated by passing a token in an orderly,
circular fashion between the LAN's peripheral devices. A token is a special pattern of bits that is circulated around the
media access domain.
The token can be modified by a LAN-attached device to form the header
of a data frame. Without this token, the data frame's header cannot be constructed
and there can be no transmission. Recipient peripheral devices copy the data
in the frame from the LAN. This device also inverts some of the bits in the
frame's header to acknowledge its receipt. The frame is then allowed to continue
traversing the ring. When it returns to its originator, that device takes
the frame off the network and strips out the data. If that device needs to
send more data, it may do so. Otherwise, the header is reconverted back into
a token and placed on the transmission media, where it travels to the next
device downstream. Although this may seem complicated, token passing is a
highly organized and efficient means of arbitrating media access permissions.
In a token-passing LAN, regardless of the media type or transmission
speed, a token-passing domain consists of all the devices that pass tokens.
Here is where a discontinuity in definitions occurs between Token Ring and
FDDI, the two most common token-passing LAN architectures. They enjoy similar
media access domain characteristics, but they define device differently.
In a Token Ring network, only the LAN's peripheral devices are counted
as devices. The hubs are nothing more than repeaters; they are incapable of
modifying token bits. Therefore, they cannot be considered devices in the
parlance of this LAN architecture. FDDI, on the other hand, does count hub ports as devices. This
has some significant ramifications on the sizes of the media access domains
in these two LAN architectures. Token Ring LANs can support up to a maximum
of 260 peripheral devices in a single token-passing domain. FDDI can support
up to 500 total devices (including hub ports) in its token-passing domain.
Therefore, you can connect more peripheral devices to a Token Ring LAN than
you can to FDDI.
Additionally, FDDI enables devices to be either single-attached (SA) or dual-attached (DA) to the LAN. These describe
the number of connections made to the LAN. DA FDDI, while providing redundancy,
effectively doubles the device count in the token-passing domain. Each attached
port must be counted as a separate device. Therefore, only 125 DA peripheral
devices may be connected to a FDDI LAN.
Many of today's LANs are constructed using repeating hubs. The result is a star-shaped physical topology;
peripheral devices are interconnected via a central hub. In other words, their
connections to the LAN radiate out from a single point much like the rays
of a star. Consequently, the physical topologies of LANs are identical, regardless
of their media access methodology.
illustrates a media access domain in a star-shaped LAN.
Given the previously described discontinuities in the definition of
the term device that exists between Ethernet, Token Ring,
and FDDI, the examples in this section apply only to Ethernet, Token Ring,
and SA FDDI.
3-1. A media access domain in a star-shaped LAN.
If a given work environment were to install a second Ethernet hub for
a second workgroup, it would have two separate Ethernet LANs. That is,
each LAN would be completely autonomous and define its own media access
and MAC broadcast
3-2 illustrates this.
3-2. Two separate LANs, each with its own media access domain.
If the two LANs in Figure
3-2 were to be
directly interconnected, the result would be a single LAN. This LAN's
media access domain would consist of all the devices that populated the
original two LANs. The media access domain would also include the two
ports used to interconnect the LANs. Therefore, this consolidated LAN
would consolidate all LAN-attached devices into a single media access
domain. This is depicted in Figure
Interconnecting the repeating hubs of LANs, both contention based and
token passing, results in a functional consolidation of their media access
domains. Depending on the LAN architecture, the expanded media access domain
may also have to include the hub ports that were interconnected. This is the
only way to expand a media access domain!
Other forms of LAN expansion and LAN-to-LAN interconnection
are possible but require additional hardware. This hardware can include LAN
switches, bridges, and routers. These devices, however, decrease the size
of the media access domain, the MAC broadcast domain, or both. Therefore,
they enable the overall size of a LAN to increase without a commensurate increase
in the size of its media access or MAC broadcast domains. Such asymmetric
expandability is the key to a LAN's scalability. These concepts are examined
throughout the remainder of this chapter.
A MAC broadcast domain
consists of all the devices connected to a LAN that receive framed data broadcast
by a machine to all other machines on the LAN. The concept of a MAC broadcast
is virtually universal throughout all IEEE-compliant LANs, regardless of their
media access methodology. Consequently, this chapter examines MAC broadcast
domains only in the context of Ethernet LANs.
FDDI is considered an IEEE-compliant LAN, even though it was not created
by the IEEE. This is because the IEEE standards are passed to the American
National Standards Institute (ANSI) for integration with their national standards.
FDDI is an ANSI specification that complies with the ANSI equivalents of the
IEEE 802.1 and 802.2 standards. Therefore, FDDI is IEEE compliant.
In essence, a MAC
broadcast domain is the set of devices that can communicate directly without
requiring higher-layer protocols or addressing. To better illustrate the
difference between MAC broadcast and media access domains, compare Figure
3-1 and Figure
3-4. An Ethernet MAC broadcast domain with five devices.
Figure 3-4 uses
the same LAN configuration depicted in Figure
3-1, but identifies its MAC
broadcast domain rather than the media access domain. The key distinction
between MAC broadcast and media access domains will become obvious as the
various LAN segmentation mechanisms are examined.
As with the media access domain, adding a second isolated LAN creates a
second, fully separate broadcast domain. Figure
3-5 identifies the MAC broadcast domains of the LAN configuration presented
in Figure 3-2.
3-5. Two separate Ethernet MAC broadcast domains.
Interconnecting these LANs in the manner demonstrated in Figure
3-3 results in a single, but larger, MAC broadcast domain. Figure
3-6 illustrates this new broadcast domain.
3-6. Making one Ethernet LAN and MAC broadcast domain of the original
This LAN's MAC broadcast domain consists of all the devices that populated
the original two LANs' broadcast domains. In this scenario, any given broadcast
message is now propagated across the network to twice as many devices as before.
Therein lies the proverbial double-edged sword of LANs with large MAC broadcast
domains: They can become quite large due to the segmentation of their media
access domains, but suffer from the flatness, or lack of a hierarchy, to their MAC broadcast domain.
LANs built with a single MAC broadcast domain are known as flat
LANs. They are flat because there is no structure or hierarchy
to their broadcast domains. The benefit of having a large broadcast domain
is that it is extremely easy to reach all the devices that are interconnected
on the LAN. The potential danger, also, is that it is extremely easy to reach
all the devices on the LAN. The more devices you connect to a flat LAN, the
more resources are consumed by each MAC broadcast message. Using the wrong
communications protocol (that is, one that makes extensive use of MAC broadcasting)
could easily compromise the performance of the network, as well as all the
devices that populate it.
MAC broadcasts are performed by setting the destination MAC address
of a frame of data to its highest possible value: FF:FF:FF:FF:FF:FF.
This reserved address value, when placed in a frame's destination address
field, is interpreted by all IEEE-compliant LANs as being addressed
to all local machines. Therefore, it is accepted by all machines, regardless
of what their actual MAC address is.
This paradox shouldn't be misinterpreted as meaning flat LANs are undesirable.
On the contrary! The introduction of LAN switching led to a flattening of
LANs. The larger a flat LAN is (in terms of its population), however, the
more important it is to segment it. Segmentation is a technique that allows
the overall size of a LAN to be expanded, by controlling the sizes of its media access and/or MAC
One of the primary architects of OpenCable, Michael
Adams, explains the key concepts of this initiative in his book
Broadband, Second Edition
by George Abe
Introduces the topics surrounding high-speed networks
to the home. It is written for anyone seeking a broad-based familiarity
with the issues of residential broadband (RBB) including product
developers, engineers, network designers, business people, professionals
in legal and regulatory positions, and industry analysts.