The Cisco Three-Layer Hierarchical Model

Most of us were exposed to hierarchy early in life. Anyone with older siblings learned what it was like to be at the bottom of the hierarchy. Regardless of where you first discovered hierarchy, today most of us experience it in many aspects of our lives. It is hierarchy that helps us understand where things belong, how things fit together, and what functions go where. It brings order and understandability to otherwise complex models. If you want a pay raise, for instance, hierarchy dictates that you ask your boss, not your subordinate. That is the person whose role it is to grant (or deny) your request. So basically, understanding hierarchy helps us discern where we should go to get what we need.

Hierarchy has many of the same benefits in network design that it does in other areas of life. When used properly, it makes networks more predictable. It helps us define which areas should perform certain functions. Likewise, you can use tools such as access lists at certain levels in hierarchical networks and avoid them at others.

Let’s face it: Large networks can be extremely complicated, with multiple protocols, detailed configurations, and diverse technologies. Hierarchy helps us summarize a complex collection of details into an understandable model. Then, as specific configurations are needed, the model dictates the appropriate manner in which to apply them.

The Cisco hierarchical model can help you design, implement, and maintain a scalable, reliable, cost-effective hierarchical internetwork. Cisco defines three layers of hierarchy, as shown in Figure 1, each with specific functions.

FIGURE 1 The Cisco hierarchical model

The following are the three layers and their typical functions:

• The core layer: backbone
• The distribution layer: routing
• The access layer: switching

Each layer has specific responsibilities. Remember, however, that the three layers are logical and are not necessarily physical devices. Consider the OSI model, another logical hierarchy. The seven layers describe functions but not necessarily protocols, right? Sometimes a protocol maps to more than one layer of the OSI model, and sometimes multiple protocols communicate within a single layer. In the same way, when we build physical implementations of hierarchical networks, we may have many devices in a single layer, or we might have a single device performing functions at two layers. The definition of the layers is logical, not physical.

Now, let’s take a closer look at each of the layers.

The Core Layer

The core layer is literally the core of the network. At the top of the hierarchy, the core layer is responsible for transporting large amounts of traffic both reliably and quickly. The only purpose of the network’s core layer is to switch traffic as fast as possible. The traffic transported across the core is common to a majority of users. However, remember that user data is processed at the distribution layer, which forwards the requests to the core if needed.

If there is a failure in the core, every single user can be affected. Therefore, fault tolerance at this layer is an issue. The core is likely to see large volumes of traffic, so speed and latency are driving concerns here. Given the function of the core, we can now consider some design specifics. Let’s start with some things we don’t want to do:
 

• Don’t do anything to slow down traffic. This includes using access lists, routing between virtual local area networks (VLANs), and implementing packet filtering.

• Don’t support workgroup access here.

• Avoid expanding the core (i.e., adding routers) when the internetwork grows. If performance becomes an issue in the core, give preference to upgrades over expansion.

Now, there are a few things that we want to do as we design the core:

• Design the core for high reliability. Consider data-link technologies that facilitate both speed and redundancy, such as FDDI, Fast Ethernet (with redundant links), or even ATM.

• Design with speed in mind. The core should have very little latency.

• Select routing protocols with lower convergence times. Fast and redundant data-link connectivity is no help if your routing tables are shot!

The Distribution Layer

The distribution layer is sometimes referred to as the workgroup layer and is the communication point between the access layer and the core. The primary functions of the distribution layer are to provide routing, filtering, and WAN access and to determine how packets can access the core, if needed. The distribution layer must determine the fastest way that network service requests are handled—for example, how a file request is forwarded to a server. After the distribution layer determines the best path, it forwards the request to the core layer if necessary.The core layer then quickly transports the request to the correct service.

The distribution layer is the place to implement policies for the network. Here you can exercise considerable flexibility in defining network operation. There are several actions that generally should be done at the distribution layer:

• Routing
• Implementing tools (such as access lists), packet filtering, and queuing
• Implementing security and network policies, including address translation and firewalls
• Redistributing between routing protocols, including static routing
• Routing between VLANs and other workgroup support functions
• Defining broadcast and multicast domains

Things to avoid at the distribution layer are limited to those functions that exclusively belong to one of the other layers.

The Access Layer

The access layer controls user and workgroup access to internetwork resources. The access layer is sometimes referred to as the desktop layer. The network resources most users need will be available locally. The distribution layer handles any traffic for remote services. The following are some of the functions to be included at the access layer:

• Continued (from distribution layer) use of access control and policies
• Creation of separate collision domains (segmentation)
• Workgroup connectivity into the distribution layer

Technologies such as DDR and Ethernet switching are frequently seen in the access layer. Static routing (instead of dynamic routing protocols) is seen here as well.

As already noted, three separate levels does not imply three separate routers. There could be fewer, or there could be more. Remember, this is a layered approach.

Data Encapsulation

When a host transmits data across a network to another device, the data goes through encapsulation: It is wrapped with protocol information at each layer of the OSI model. Each layer communicates only with its peer layer on the receiving device.

To communicate and exchange information, each layer uses Protocol Data Units (PDUs). These hold the control information attached to the data at each layer of the model. They are usually attached to the header in front of the data field but can also be in the trailer, or end, of it.

Each PDU attaches to the data by encapsulating it at each layer of the OSI model, and each has a specific name depending on the information provided in each header. This PDU information is read only by the peer layer on the receiving device. After it’s read, it’s stripped off and the data is then handed to the next layer up.

Figure 1 shows the PDUs and how they attach control information to each layer. This figure demonstrates how the upper-layer user data is converted for transmission on the network. The data stream is then handed down to the Transport layer, which sets up a virtual circuit to the receiving device by sending over a synch packet. Next, the data stream is broken up into smaller pieces, and a Transport layer header (a PDU) is created and attached to the header of the data field; now the piece of data is called a segment. Each segment is sequenced so the data stream can be put back together on the receiving side exactly as it was transmitted.

Each segment is then handed to the Network layer for network addressing and routing through the internetwork. Logical addressing (for example, IP) is used to get each segment to the correct network. The Network layer protocol adds a control header to the segment handed down from the Transport layer, and what we have now is called a packet or datagram. Remember that the Transport and Network layers work together to rebuild a data stream on a receiving host, but it’s not part of their work to place their PDUs on a local network segment—which is the only way to get the information to a router or host.

It’s the Data Link layer that’s responsible for taking packets from the Network layer and placing them on the network medium (cable or wireless). The Data Link layer encapsulates each packet in a frame, and the frame’s header carries the hardware address of the source and destination hosts. If the destination device is on a remote network, then the frame is sent to a router to be routed through an internetwork. Once it gets to the destination network, a new frame is used to get the packet to the destination host.

To put this frame on the network, it must first be put into a digital signal. Since a frame is really a logical group of 1s and 0s, the Physical layer is responsible for encoding these digits into a digital signal, which is read by devices on the same local network. The receiving devices will synchronize on the digital signal and extract (decode) the 1s and 0s from the digital signal. At this point, the devices build the frames, run a CRC, and then check their answer against the answer in the frame’s FCS field. If it matches, the packet is pulled from the frame and what’s left of the frame is discarded. This process is called de-encapsulation.

The packet is handed to the Network layer, where the address is checked. If the address matches, the segment is pulled from the packet and what’s left of the packet is discarded. The segment is processed at the Transport layer, which rebuilds the data stream and acknowledges to the transmitting station that it received each piece. It then happily hands the data stream to the upper-layer application.

At a transmitting device, the data encapsulation method works like this:

1. User information is converted to data for transmission on the network.
2. Data is converted to segments and a reliable connection is set up between the transmitting
and receiving hosts.
3. Segments are converted to packets or datagrams, and a logical address is placed in the
header so each packet can be routed through an internetwork.
4. Packets or datagrams are converted to frames for transmission on the local network. Hardware
(Ethernet) addresses are used to uniquely identify hosts on a local network segment.
5. Frames are converted to bits, and a digital encoding and clocking scheme is used.
6. To explain this in more detail using the layer addressing, I’ll use Figure 2.

FIGURE 2 PDU and layer addressing

Remember that a data stream is handed down from the upper layer to the Transport layer. As technicians, we really don’t care who the data stream comes from because that’s really a programmer’s problem. Our job is to rebuild the data stream reliably and hand it to the upper layers on the receiving device.

Before we go further in our discussion of Figure 2, let’s discuss port numbers and make sure we understand them. The Transport layer uses port numbers to define both the virtual circuit and the upper-layer process, as you can see from Figure 3.

FIGURE 3 Port numbers at the Transport layer

The Transport layer takes the data stream, makes segments out of it, and establishes a reliable session by creating a virtual circuit. It then sequences (numbers) each segment and uses acknowledgments and flow control. If you’re using TCP, the virtual circuit is defined by the source port number. Remember, the host just makes this up starting at port number 1024 (0 through 1023 are reserved for well-known port numbers). The destination port number defines the upper-layer process (application) that the data stream is handed to when the data stream is reliably rebuilt on the receiving host.

Now that you understand port numbers and how they are used at the Transport layer, let’s go back to Figure 3. Once the Transport layer header information is added to the piece of data, it becomes a segment and is handed down to the Network layer along with the destination IP address. (The destination IP address was handed down from the upper layers to the Transport layer with the data stream, and it was discovered through a name resolution method at the upper layers—probably DNS.)

The Network layer adds a header, and adds the logical addressing (IP addresses), to the front of each segment. Once the header is added to the segment, the PDU is called a packet. The packet has a protocol field that describes where the segment came from (either UDP or TCP) so it can hand the segment to the correct protocol at the Transport layer when it reaches the receiving host.

The Network layer is responsible for finding the destination hardware address that dictates where the packet should be sent on the local network. It does this by using the Address Resolution Protocol (ARP). IP at the Network layer looks at the destination IP address and compares that address to its own source IP address and subnet mask. If it turns out to be a local network request, the hardware address of the local host is requested via an ARP request. If the packet is destined for a remote host, IP will look for the IP address of the default gateway (router) instead.

The packet, along with the destination hardware address of either the local host or default gateway, is then handed down to the Data Link layer. The Data Link layer will add a header to the front of the packet and the piece of data then becomes a frame. (We call it a frame because both a header and a trailer are added to the packet, which makes the data resemble bookends or a frame, if you will.) This is shown in Figure 2. The frame uses an Ether-Type field to describe which protocol the packet came from at the Network layer. Now a cyclic redundancy check (CRC) is run on the frame, and the answer to the CRC is placed in the Frame Check Sequence field found in the trailer of the frame.

The frame is now ready to be handed down, one bit at a time, to the Physical layer, which will use bit timing rules to encode the data in a digital signal. Every device on the network segment will synchronize with the clock and extract the 1s and 0s from the digital signal and build a frame. After the frame is rebuilt, a CRC is run to make sure the frame is okay. If everything turns out to be all good, the hosts will check the destination address to see if the frame is for them. If all this is making your eyes cross and your brain freeze, don’t freak.

Ethernet Cabling

Ethernet cabling is an important discussion, especially if you are planning on taking the Cisco exams. Three types of Ethernet cables are available:

• Straight-through cable
• Crossover cable
• Rolled cable

We will look at each in the following sections.

Straight-Through Cable

The straight-through cable is used to connect

• Host to switch or hub
• Router to switch or hub

Four wires are used in straight-through cable to connect Ethernet devices. It is relatively simple to create this type; Figure 1 shows the four wires used in a straight-through Ethernet cable.

Notice that only pins 1, 2, 3, and 6 are used. Just connect 1 to 1, 2 to 2, 3 to 3, and 6 to 6 and you’ll be up and networking in no time. However, remember that this would be an Ethernet-only cable and wouldn’t work with voice, Token Ring, ISDN, and so on.

FIGURE 1 Straight-through Ethernet cable

Crossover Cable

The crossover cable can be used to connect

• Switch to switch
• Hub to hub
• Host to host
• Hub to switch
• Router direct to host

The same four wires are used in this cable as in the straight-through cable; we just connect different pins together. Figure 2 shows how the four wires are used in a crossover Ethernet cable.

Notice that instead of connecting 1 to 1, 2 to 2, and so on, here we connect pins 1 to 3 and 2 to 6 on each side of the cable.

FIGURE 2 Crossover Ethernet cable

Rolled Cable

Although rolled cable isn’t used to connect any Ethernet connections together, you can use a rolled Ethernet cable to connect a host to a router console serial communication (com) port.

If you have a Cisco router or switch, you would use this cable to connect your PC running HyperTerminal to the Cisco hardware. Eight wires are used in this cable to connect serial devices, although not all eight are used to send information, just as in Ethernet networking.

Figure 3 shows the eight wires used in a rolled cable.

FIGURE 3 Rolled Ethernet cable

These are probably the easiest cables to make because you just cut the end off on one side of a straight-through cable, turn it over, and put it back on (with a new connector, of course).

Once you have the correct cable connected from your PC to the Cisco router or switch, you can start HyperTerminal to create a console connection and configure the device. Set the configuration as follows:

1. Open HyperTerminal and enter a name for the connection. It is irrelevant what you name it, but I always just use Cisco. Then click OK.

2. Choose the communications port—either COM1 or COM2, whichever is open on your PC.

3. Now set the port settings. The default values (2400bps and no flow control hardware) will not work; you must set the port settings as shown in Figure 4.

Notice that the bit rate is now set to 9600 and the flow control is set to None. At this point, you can click OK and press the Enter key and you should be connected to your Cisco device console port.

We’ve taken a look at the various RJ45 unshielded twisted pair (UTP) cables. Keeping this in mind, what cable is used between the switches in Figure 5? In order for host A to ping host B, you need a crossover cable to connect the two switches together. But what types of cables are used in the network shown in Figure 6? In Figure 6, there are a variety of cables in use. For the connection between the switches, we’d obviously use a crossover cable like we saw in Figure 1. The trouble is, we have a console connection that uses a rolled cable. Plus, the connection from the router to the switch is a straight-through cable, as is true for the hosts to the switches. Keep in mind that if we had a serial connection (which we don’t), it would be a V.35 that we’d use to connect us to a WAN.

FIGURE 4 Port settings for a rolled cable connection

FIGURE 5 RJ45 UTP cable question #1

FIGURE 6 RJ45 UTP cable question #2

Ethernet at the Physical Layer

Ethernet was first implemented by a group called DIX (Digital, Intel, and Xerox). They created and implemented the first Ethernet LAN specification, which the IEEE used to create the IEEE 802.3 Committee. This was a 10Mbps network that ran on coax and then eventually twisted pair and fiber physical media.

The IEEE extended the 802.3 Committee to two new committees known as 802.3u (Fast Ethernet) and 802.3ab (Gigabit Ethernet on category 5) and then finally 802.3ae (10Gbps over fiber and coax).

Figure 1 shows the IEEE 802.3 and original Ethernet Physical layer specifications.

When designing your LAN, it’s really important to understand the different types of Ethernet media available to you. Sure, it would be great to run Gigabit Ethernet to each desktop and 10Gbps between switches, and although this might happen one day, justifying the cost of that network today would be pretty difficult. But if you mix and match the different types of Ethernet media methods currently available, you can come up with a cost-effective network solution that works great.

FIGURE 1 Ethernet Physical layer specifications

The EIA/TIA (Electronic Industries Association and the newer Telecommunications Industry Alliance) is the standards body that creates the Physical layer specifications for Ethernet.The EIA/TIA specifies that Ethernet use a registered jack (RJ) connector with a 4 5 wiring sequence on unshielded twisted-pair (UTP) cabling (RJ45). However, the industry is moving toward calling this just an 8-pin modular connector.

Each Ethernet cable type that is specified by the EIA/TIA has inherent attenuation, which is defined as the loss of signal strength as it travels the length of a cable and is measured in decibels (dB). The cabling used in corporate and home markets is measured in categories. A higherquality cable will have a higher-rated category and lower attenuation. For example, category 5 is better than category 3 because category 5 cables have more wire twists per foot and therefore less crosstalk. Crosstalk is the unwanted signal interference from adjacent pairs in the cable.

Here are the original IEEE 802.3 standards:

10Base2 10Mbps, baseband technology, up to 185 meters in length. Known as thinnet and
can support up to 30 workstations on a single segment. Uses a physical and logical bus with
AUI connectors. The 10 means 10Mbps, Base means baseband technology (which is a signaling
method for communication on the network), and the 2 means almost 200 meters. 10Base2
Ethernet cards use BNC (British Naval Connector, Bayonet Neill Concelman, or Bayonet Nut
Connector) and T-connectors to connect to a network.

10Base5 10Mbps, baseband technology, up to 500 meters in length. Known as thicknet.
Uses a physical and logical bus with AUI connectors. Up to 2,500 meters with repeaters and
1,024 users for all segments.

10BaseT 10Mbps using category 3 UTP wiring. Unlike with the 10Base2 and 10Base5 networks,
each device must connect into a hub or switch, and you can have only one host per segment
or wire. Uses an RJ45 connector (8-pin modular connector) with a physical star topology
and a logical bus.

Each of the 802.3 standards defines an Attachment Unit Interface (AUI), which allows a one-bit-at-a-time transfer to the Physical layer from the Data Link media access method. This allows the MAC to remain constant but means the Physical layer can support any existing and new technologies. The original AUI interface was a 15-pin connector, which allowed a transceiver(transmitter/receiver) that provided a 15-pin-to-twisted-pair conversion.

The thing is, the AUI interface cannot support 100Mbps Ethernet because of the high frequencies involved. So 100BaseT needed a new interface, and the 802.3u specifications created one called the Media Independent Interface (MII), which provides 100Mbps throughput. The MII uses a nibble, defined as 4 bits. Gigabit Ethernet uses a Gigabit Media Independent Interface(GMII) and transmits 8 bits at a time.

802.3u (Fast Ethernet) is compatible with 802.3 Ethernet because they share the same physical characteristics. Fast Ethernet and Ethernet use the same maximum transmission unit (MTU), use the same MAC mechanisms, and preserve the frame format that is used by 10BaseT Ethernet.

Basically,Fast Ethernet is just based on an extension to the IEEE 802.3 specification, except that it
offers a speed increase of 10 times that of 10BaseT.

Here are the expanded IEEE Ethernet 802.3 standards:

100BaseTX (IEEE 802.3u) EIA/TIA category 5, 6, or 7 UTP two-pair wiring. One user per segment; up to 100 meters long. It uses an RJ45 connector with a physical star topology and a logical bus.

100BaseFX (IEEE 802.3u) Uses fiber cabling 62.5/125-micron multimode fiber. Point to point topology; up to 412 meters long. It uses an ST or SC connector, which are media interface connectors.

1000BaseCX (IEEE 802.3z) Copper twisted-pair called twinax (a balanced coaxial pair) that can only run up to 25 meters.

1000BaseT (IEEE 802.3ab) Category 5, four-pair UTP wiring up to 100 meters long.

1000BaseSX (IEEE 802.3z) MMF using 62.5- and 50-micron core; uses an 850 nano-meter laser and can go up to 220 meters with 62.5-micron, 550 meters with 50-micron.

1000BaseLX (IEEE 802.3z) Single-mode fiber that uses a 9-micron core and 1300 nanometer laser and can go from 3 kilometers up to 10 kilometers.

Note:If you want to implement a network medium that is not susceptible to electromagnetic interference (EMI), fiber-optic cable provides a more secure, long-distance cable that is not susceptible to EMI at high speeds.

Ethernet at the Data Link Layer

Ethernet at the Data Link layer is responsible for Ethernet addressing, commonly referred to as hardware addressing or MAC addressing. Ethernet is also responsible for framing packets received from the Network layer and preparing them for transmission on the local network through the Ethernet contention media access method.

Ethernet Addressing

Here’s where we get into how Ethernet addressing works. It uses the Media Access Control(MAC) address burned into each and every Ethernet network interface card (NIC). The MAC, or hardware, address is a 48-bit (6-byte) address written in a hexadecimal format.

Figure 1 shows the 48-bit MAC addresses and how the bits are divided.

FIGURE 1 Ethernet addressing using MAC addresses

The organizationally unique identifier (OUI) is assigned by the IEEE to an organization.

It’s composed of 24 bits, or 3 bytes. The organization, in turn, assigns a globally administered address (24 bits, or 3 bytes) that is unique (supposedly, again—no guarantees) to each and every adapter it manufactures. Look closely at the figure. The high-order bit is the Individual/ Group (I/G) bit. When it has a value of 0, we can assume that the address is the MAC address of a device and may well appear in the source portion of the MAC header. When it is a 1, we can assume that the address represents either a broadcast or multicast address in Ethernet or a broadcast or functional address in TR and FDDI (who really knows about FDDI?).

The next bit is the global/local bit, or just G/L bit (also known as U/L, where U means universal). When set to 0, this bit represents a globally administered address (as by the IEEE). When the bit is a 1, it represents a locally governed and administered address (as in what DECnet used to do).

The low-order 24 bits of an Ethernet address represent a locally administered or manufacturer assigned code. This portion commonly starts with 24 0s for the first card made and continues in order until there are 24 1s for the last (16,777,216th) card made. You’ll find that many manufacturers use these same six hex digits as the last six characters of their serial number on the same card.

Ethernet Frames

The Data Link layer is responsible for combining bits into bytes and bytes into frames. Frames are used at the Data Link layer to encapsulate packets handed down from the Network layer for transmission on a type of media access.

The function of Ethernet stations is to pass data frames between each other using a group of bits known as a MAC frame format. This provides error detection from a cyclic redundancy check (CRC). But remember—this is error detection, not error correction. The 802.3 frames and Ethernet frame are shown in Figure 2

Note:Encapsulating a frame within a different type of frame is called tunneling.

FIGURE 2  802.3 and Ethernet frame formats

Following are the details of the different fields in the 802.3 and Ethernet frame types:

Preamble An alternating 1,0 pattern provides a 5MHz clock at the start of each packet, which allows the receiving devices to lock the incoming bit stream.

Start Frame Delimiter (SFD)/Synch The preamble is seven octets and the SFD is one octet(synch). The SFD is 10101011, where the last pair of 1s allows the receiver to come into the alternating 1,0 pattern somewhere in the middle and still sync up and detect the beginning of the data.

Destination Address (DA) This transmits a 48-bit value using the least significant bit(LSB) first. The DA is used by receiving stations to determine whether an incoming packet is addressed to a particular node. The destination address can be an individual address or a broadcast or multicast MAC address. Remember that a broadcast is all 1s (or Fs in hex) and is sent to all devices but a multicast is sent only to a similar subset of nodes on a network.

Source Address (SA) The SA is a 48-bit MAC address used to identify the transmitting device, and it uses the LSB first. Broadcast and multicast address formats are illegal within the SA field.

Length or Type 802.3 uses a Length field, but the Ethernet frame uses a Type field to identify the Network layer protocol. 802.3 cannot identify the upper-layer protocol and must be used with a proprietary LAN—IPX, for example.

Data This is a packet sent down to the Data Link layer from the Network layer. The size can vary from 64 to 1,500 bytes.

Frame Check Sequence (FCS) FCS is a field at the end of the frame that’s used to store the CRC.

Let’s pause here for a minute and take a look at some frames caught on our trusty OmniPeek network analyzer. You can see that the frame below has only three fields: Destination, Source, and Type (shown as Protocol Type on this analyzer):

Destination: 00:60:f5:00:1f:27
Source: 00:60:f5:00:1f:2c
Protocol Type: 08-00 IP

This is an Ethernet_II frame. Notice that the type field is IP, or 08-00 (mostly just referred to as 0x800) in hexadecimal.

The next frame has the same fields, so it must be an Ethernet_II frame too:

Destination: ff:ff:ff:ff:ff:ff Ethernet Broadcast

Source: 02:07:01:22:de:a4
Protocol Type: 08-00 IP

Did you notice that this frame was a broadcast? You can tell because the destination hardware address is all 1s in binary, or all Fs in hexadecimal.

Let’s take a look at one more Ethernet_II frame. I’ll talk about this next example again when we use IPv6 in this blog, but you can see that the Ethernet frame is the same Ethernet_II frame we use with the IPv4 routed protocol but the type field has 0x86dd when we are carrying IPv6 data, and when we have IPv4 data, we use 0x0800 in the protocol field:

Destination: IPv6-Neighbor-Discovery_00:01:00:03 (33:33:00:01:00:03)
Source: Aopen_3e:7f:dd (00:01:80:3e:7f:dd)
Type: IPv6 (0x86dd)

This is the beauty of the Ethernet_II frame. Because of the protocol field, we can run any Network layer routed protocol and it will carry the data because it can identify the Network layer protocol.

Half- and Full-Duplex Ethernet

Half-duplex Ethernet is defined in the original 802.3 Ethernet; Cisco says it uses only one wire pair with a digital signal running in both directions on the wire. Certainly, the IEEE specifications discuss the process of half duplex somewhat differently, but what Cisco is talking about is a general sense of what is happening here with Ethernet.

It also uses the CSMA/CD protocol to help prevent collisions and to permit retransmitting if a collision does occur. If a hub is attached to a switch, it must operate in half-duplex mode because the end stations must be able to detect collisions. Half-duplex Ethernet—typically 10BaseT—is only about 30 to 40 percent efficient as Cisco sees it because a large 10BaseT network will usually only give you 3 to 4Mbps, at most.

But full-duplex Ethernet uses two pairs of wires instead of one wire pair like half duplex. And full duplex uses a point-to-point connection between the transmitter of the transmitting device and the receiver of the receiving device. This means that with full-duplex data transfer, you get a faster data transfer compared to half duplex. And because the transmitted data is sent on a different set of wires than the received data, no collisions will occur.

The reason you don’t need to worry about collisions is because now it’s like a freeway with multiple lanes instead of the single-lane road provided by half duplex. Full-duplex Ethernet is supposed to offer 100 percent efficiency in both directions—for example, you can get 20Mbps with a 10Mbps Ethernet running full duplex or 200Mbps for Fast Ethernet. But this rate is something known as an aggregate rate, which translates as “you’re supposed to get” 100 percent efficiency. No guarantees, in networking as in life.

Full-duplex Ethernet can be used in three situations:

• With a connection from a switch to a host
• With a connection from a switch to a switch
• With a connection from a host to a host using a crossover cable

Note: Full-duplex Ethernet requires a point-to-point connection when only two nodes are present. You can run full-duplex with just about any device except a hub.

Now, if it’s capable of all that speed, why wouldn’t it deliver? Well, when a full-duplex Ethernet port is powered on, it first connects to the remote end and then negotiates with the other end of the Fast Ethernet link. This is called an auto-detect mechanism. This mechanism first decides on the exchange capability, which means it checks to see if it can run at 10 or 100Mbps. It then checks to see if it can run full duplex, and if it can’t, it will run half duplex.

Note: Remember that half-duplex Ethernet shares a collision domain and provides a lower effective throughput than full-duplex Ethernet, which typically has a private collision domain and a higher effective throughput.

Lastly, remember these important points:

• There are no collisions in full-duplex mode.
• A dedicated switch port is required for each full-duplex node.
• The host network card and the switch port must be capable of operating in full-duplex mode.
• Now let’s take a look at how Ethernet works at the Data Link layer.

Ethernet Networking uses CSMA/CD protocol

Ethernet is a contention media access method that allows all hosts on a network to share the same bandwidth of a link. Ethernet is popular because it’s readily scalable, meaning that it’s comparatively easy to integrate new technologies, such as Fast Ethernet and Gigabit Ethernet, into an existing network infrastructure. It’s also relatively simple to implement in the first place, and with it, troubleshooting is reasonably straightforward. Ethernet uses both Data Link and Physical layer specifications, and this section of the chapter will give you both the Data Link layer and Physical layer information you need to effectively implement, troubleshoot, and maintain an Ethernet network.

Ethernet networking uses Carrier Sense Multiple Access with Collision Detection (CSMA/CD), a protocol that helps devices share the bandwidth evenly without having two devices transmit at the same time on the network medium. CSMA/CD was created to overcome the problem of those collisions that occur when packets are transmitted simultaneously from different nodes. And trust me—good collision management is crucial, because when a node transmits in a CSMA/CD network, all the other nodes on the network receive and examine that transmission. Only bridges and routers can effectively prevent a transmission from propagating throughout the entire network!

So, how does the CSMA/CD protocol work? Let’s start by taking a look at Figure 1

FIGURE 1 CSMA/CD

When a host wants to transmit over the network, it first checks for the presence of a digital signal on the wire. If all is clear (no other host is transmitting), the host will then proceed with its transmission. But it doesn’t stop there. The transmitting host constantly monitors the wire to make sure no other hosts begin transmitting. If the host detects another signal on the wire, it sends out an extended jam signal that causes all nodes on the segment to stop sending data (think busy signal). The nodes respond to that jam signal by waiting a while before attempting to transmit again. Backoff algorithms determine when the colliding stations can retransmit. If collisions keep occurring after 15 tries, the nodes attempting to transmit will then timeout. Pretty clean!

When a collision occurs on an Ethernet LAN, the following happens:

• A jam signal informs all devices that a collision occurred.
• The collision invokes a random backoff algorithm.
• Each device on the Ethernet segment stops transmitting for a short time until the timers expire.
• All hosts have equal priority to transmit after the timers have expired.

The following are the effects of having a CSMA/CD network sustaining heavy collisions:

• Delay
• Low throughput
• Congestion