Softpanorama (slightly skeptical) Open Source Software Educational Society

May the source be with you, but remember the KISS principle ;-)

Ethernet Protocol

(Adapted from Lecture notes for FDU students by Professor Nikolai Bezroukov)

Summary

Introduction

CSMA/CD Access Method

Half-duplex and Full Duplex Mode

Ethernet Statistics

• Collision rates

• Input/Output Errors

Ethernet Addresses

• Setting a Local Ethernet Address

Ethernet frame format

• IEEE 802.3 frame format
• Ethernet version 2 frame format

Maximum Transfer Units

Ethernet Frame Errors

IP Configuration

Troubleshooting

Summary

• Carrier Sense Multiple Access With Collision Detection (CSMA/CD) is a network control protocol in which
  • a carrier sensing scheme is used (each interface monitors the network for the a carrier signal (Carrier Sense))
  • During the gap between transmissions, each interface has an equal chance to transmit data
  • If a transmitting station detects another signal while transmitting a frame (Collision Detection that is performed in transceiver circuitry) , it stops transmitting that frame, transmits a jam signal, and then waits for a random time interval (known as "backoff delay" and determined using the truncated binary exponential backoff algorithm) before trying to send that frame again.
     
• A duplex communication system is a system composed of two connected parties or devices which can communicate with one another in both directions.
  • A half-duplex system provides for communication in both directions, but only one direction at a time (not simultaneously). Typically, once a party begins receiving a signal, it must wait for the transmitter to stop transmitting, before replying.
  • A full-duplex system allows communication in both directions, and unlike half-duplex, allows this to happen simultaneously.
     
• Netstat command displays important Ethernet statistics as well as important IP statistic and parameters (routing table, active connection, etc).   With option - i  it displays statistics  including Ethernet statistics like MTU (Solaris). Used without parameters, netstat displays active TCP connections.

  netstat [-a] [-i] [ -e] [-n] [-o] [-p Protocol] [-r] [-s] [Interval]

   

• Collision rates.  A collision is the mechanism used by Ethernet to control access and allocate shared bandwidth among stations that want to transmit at the same time on a shared medium. Because the medium is shared, a mechanism must exist where two stations can detect that they want to transmit at the same time. This mechanism is collision detection. Collision rates higher then 5% on 10-Mbps LAN and 10% on 100-Mbps LAN are first indications of the network overload. Switches minimize collisions so this is not a problem in modern LAN.
   
• An Ethernet address is sometimes referred to as a media access control (MAC) address. An Ethernet address is 48 bits long and is displayed as 12 hexadecimal digits (six groups of two digits) separated by colons. An example of an Ethernet address is  44-45-53-54-42-00
  • The IEEE administers unique Ethernet addresses. IEEE designates the first three octets as vendor-specific.
  • There are three types of addresses: unicast, broadcast, and multicast.
    • Unicast Addresses. Unicast addresses are used for one-to-one communication. The system uses a unicast address to send a message to another system on the local Ethernet network. You can use a system’s unique Ethernet address as a unicast address.
    • Broadcast Addresses. A device uses a broadcast address to send messages to all systems on the local Ethernet network. The Ethernet broadcast address is represented in the form of all 1s in binary format and as ff:ff:ff:ff:ff:ff in hexadecimal format. When the Network Interface layer receives an Ethernet frame with a destination address of all 1s, it passes the address to the next layer for processing.
    • Multicast Addresses. A system uses a multicast address to send a message to a subset of systems on the local Ethernet. In Ethernet multicast addressing, the value of the first three octets determines if the address is multicast. The last three octets determine the specific multicast’s group identity.
  • To view the current host-based Ethernet address, perform the command at the ok prompt:

    ok banner
     

  • To display the Ethernet address assigned to each interface, perform the command:

    # ifconfig -a

    
    hme0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
    inet 192.168.30.31 netmask ffffff00 broadcast 192.168.30.255
    ether 8:0:20:b9:72:23
     

  • You need to set the local-mac-address? variable in the system’s EEPROM to true to be able to modify MAC address of the adapters in the system.  To view the contents of the EEPROM for the definition of the local-mac-address? variable, perform the command:

    # eeprom local-mac-address?
    local-mac-address?=false

• The maximum transfer unit (MTU) is the largest amount of data that can be transferred across a physical network. The Ethernet MTU is hardware  specific. For a physical Ethernet interface, the MTU is 1500 bytes, while the MTU is 8232 bytes for a loopback interface.
   
• Ethernet Frame Errors are:
  • Runts Packets that are less than 64 bytes are too short and are discarded. Runts are usually caused by collisions. These can be formed by poor wiring and electrical interference.
  • Jabbers Packets that are greater than 1500 bytes (MTU) are too long and are discarded. These indicate that a device has electrical problems.
  • Long A frame that is between 1518 and 6000 bytes long, often because of faulty hardware or software on the sending system.
  • Giant A frame that is more than 6000 bytes long, often because of faulty hardware or software on the sending system.
  • Bad CRC If the received packet fails the CRC, the packet is corrupted and discarded. This is also known as a frame check sequence (FCS) error.
     

• Troubleshooting

  • /sbin/ifconfig:  An ifconfig -a will give you information about all NIC, including IP address (inet), subnet mask (netmask), and if run as root, the MAC address (ether). Only root gets full information from this command.

  • Netatat

    • netstat -i # General information with statistics on each NIC. Often used in homemade monitoring scripts.
    • netstat -a. # Port/Socket information can be had with a
    • netstat -rn # most popular way to check current system routing tables with .
  • /usr/sbin/snoop
    • snoop -d hme0 broadcast #only broadcast frames in summary mode
    • snoop -d qfe0 -V 192.168.1.2
    • snoop -d qfe0 -o /tmp/snooper 192.168.1.2 # capture to a file
    • snoop -d qfe0 -o /tmp/snooper broadcast # same with broadcast
    • snoop -i /tmp/snooper -V | egrep -iv 'nfs|ack|contin|ftp|ip'
  • /usr/sbin/arp: arp allows you to view that table to see what kind of information your computer has cached, as well as to input or delete entries, in case of problems. When having DNS issues, an arp command with the -an option will list the entire table without resolving IP address to host names. arp is one way to get the MAC address of the NIC answering for a host.

Introduction

At its most basic form, a network consists of two computers that connect to each other to collaborate on a particular task. These collaborative tasks range from simple file transfers to the complex tasks of distributed computing or clustering.

There are two basic parts to a network: the physical medium and the network protocol. The physical medium employed in a network varies, depending on how you need to connect your computers, and how much you are willing to spend. Some of the common methods used for computer connectivity include:

Ethernet is a baseband networking technology that has been defined by various standards. Ethernet is the most dominant LAN technology:

A baseband technology devotes the entire bandwidth of the media to one channel. Simplistically, bandwidth can be defined as a measure of network capacity - the greater the available bandwidth, the greater the network's resilience to loading. The inclusion of the word "base" in 10BaseT, 10Base5 etc., simply refers to Ethernet's baseband nature. The most common IEEE media identifiers are:

• 10BaseT This is older standard office/home networking technology that uses cables with wires that are twisted in pairs. A hub/switch allows for the connection of multiple computers. It uses unshielded twisted pair category 5 cabling  and RJ45 connectors. UTP cable resembles domestic telephone wire. The RJ45 connectors are American telephone connectors. 10BaseT operates at 10Mbps or 100 Mbps with a maximum 100 meters segment length. 10BaseT uses a star physical topology.
   

• 100BASE-TX Media Type The 100BASE-TX media type is based on specifications published in the American National Standards Institute (ANSI) Twisted-Pair – Physical Media Standard (TP-PMD). The 100BASE-TX media type carries 100 Mbps signals over two pairs of wire. Because the ANSI TP-PMD specification provides for the use of either unshielded twisted-pair or shielded twisted-pair cable, 100BASE-TX uses both. Requres Category 5 cable.
   
• 100BASE-T4 Media Type The 100BASE-T4 media type operates over four pairs of wires. The
  signaling system makes it possible to provide fast Ethernet signals (100 MHz) over any existing standard voice-grade Category 3 or 4 unshielded twisted-pair cable that might be installed. One pair of wires transmits data (TX), one pair receives data (RX), and two pairs are bidirectional (BI) data pairs. The 100BASE-T4 specifications recommend using Category 5 patch cables, jumpers, and connecting hardware whenever possible because these higher-quality components and cables improve the reception of signals on the link.
   
• 100BASE-FX Media Type The 100BASE-FX (fast fiber-optic) media system uses pulses of light instead of electrical currents to send signals. The use of fiber provides superior electrical isolation for equipment at each end of the fiber link. While LAN equipment used in metallic media segments has protection circuits designed for typical indoor electrical hazards, fiber-optic media is nonconductive. This complete electrical isolation provides immunity from much larger electrical hazards, such as lightning strikes, and from the flow of current that can result from having different levels of electrical ground currents that can be found in separate buildings. Complete electrical isolation is essential when using LAN segments to link separate buildings.

  An advantage of the 100BASE-FX fiber-optic link segment is that it can span long distances. Fiber also provides more security because the optical signal does not cause induction.
   

• 1000BASE-X Media Type In 1998, the IEEE Standards Board approved 802.3z, the gigabit Ethernet standard for 1000 Mbps over multimode fiber (MMF) and single-mode fiber (SMF). Gigabit Ethernet is an extension of the successful 10-Mbps and 100-Mbps 802.3 standards. Gigabit Ethernet provides a raw bandwidth of 1000 Mbps and maintains full compatibility with the installed base of over 100 million Ethernet nodes. Gigabit Ethernet includes both full-duplex and half-duplex operating modes. The 1000BASE-X standard refers to two implementations of fiber-optic segment types: 1000BASE-SX and 1000BASE-LX. 1000BASE-SX Media Type:
   
  • 1000BASE-SX media system is the shortest wavelength specification because it uses short wavelength lasers to transmit data over fiber-optic cable. Sun’s implementation of the 1000BASE-SX system specification supports the following distances:
     
    • 300 meters over 62.5-micron MMF cable
    • 550 meters over 50-micron MMF cable 1000BASE-LX Media Type
       
  • 1000BASE-LX media system (gigabit Ethernet over Multi-mode Fiber and Single-mode Fiber) is the longest wavelength specification because it uses longwave lasers to transmit data over fiber-optic cable. Sun’s implementation of the 1000BASE-SX system specification supports the following distances:
    • 550 meters over 62.5-micron and 50-micron MMF cable
    • 3000 meters over 9-micron SMF cable

Ethernet was designed as a packet-switching LAN over broadcast technology. Devices connect to the network and compete for access to a shared communications channel. The IEEE 802.3 standard for Ethernet was defined in 1985. Ethernet standards are implemented at the Network Interface layer of the TCP/IP protocol model. The three major elements of Ethernet networks are:

• Ethernet packets are called frames. These are units of data sent across the network.
• The Ethernet access method, CSMA/CD. This method controls packet transmission and information flow across the Ethernet hardware.
• Hardware cables, connectors, and circuitry. These transfer data to and from systems across the network.

There are no levels of priority for transmission. All computers on an Ethernet network are considered equal in terms of their access to the media. Unless it is switched :

• Ethernet uses carrier sensed multiple access with collision detection (CSMA/CD) as its method of medium access control.
• Data in the Ethernet are accessible to all connected devices.
• High load on a LAN segment can lead to packet collisions and a marked drop in network efficiency.
• Ethernet has a broadcast logical topology - data is sent to all nodes on the network simultaneously.
• Ethernet remains the most common LAN architecture. ATM has become important as a LAN technology in areas of the computing industry where high data transfer rates are required. Multimedia, computer graphics and video conferencing all need fast networks, therefore ATM is a good choice, but the emergent gigabit Ethernet may challenge ATM.  
• Ethernet corresponds to the network interface layer of TCP/IP model.
  
• Interfaces have been developed to connect all types of computer, across many types of media, to Ethernet. PCs use a network interface card (NIC) with RJ45 connector .

Even if two networks are using the same medium, they can be using two completely different protocols or methods of communication. Some protocols used in today's networks include:

• TCP/IP -- The Internet standard. Each NIC is assigned an IP address and subnet mask. These are used to determine which machines are in the local network and which require traffic to be sent through a router.
• IPX/SPX -- A broadcast protocol developed by Novell. Each NIC is assigned a unique IPX network address.
• AppleTalk -- An Apple proprietary protocol. Each NIC is assigned a network number, a node number, and a socket number. Similar to TCP/IP in its configuration and routing abilities.

CSMA/CD Access Method

Partly derived from Federal Standard 1037C and Wikipedia Article

Carrier Sense Multiple Access With Collision Detection (CSMA/CD) is a network control protocol in which

• a carrier sensing scheme is used (each interface monitors the network for the a carrier signal (Carrier Sense))
• During the gap between transmissions, each interface has an equal chance to transmit data
• If a transmitting station detects another signal while transmitting a frame (Collision Detection that is performed in transceiver circuitry) , it stops transmitting that frame, transmits a jam signal, and then waits for a random time interval (known as "backoff delay" and determined using the truncated binary exponential backoff algorithm) before trying to send that frame again.

CSMA/CD is a modification of pure Carrier Sense Multiple Access (CSMA). In CSMA/CD, sending nodes are able to detect when a collision occurs and stop transmitting immediately, backing off for a random amount of time before trying again. This results in much more efficient use of the media since the bandwidth of transmitting the entire frame is not wasted. However, it is not possible with all media (e.g., radio), and requires extra electronics (not an issue with today's technology, but one reason Apple used CSMA/CA-based LocalTalk instead of the then much more expensive Ethernet).

Collision detection is used to improve CSMA performance by terminating transmission as soon as a collision is detected, and reducing the probability of a second collision on retry.

Methods for collision detection are media dependent, but on an electrical bus such as Ethernet, collisions can be detected by comparing transmitted data with received data. If they differ, another transmitter is overlaying the first transmitter's signal (a collision), and transmission terminates immediately. A jam signal is sent which will cause all transmitters to back off by random intervals, reducing the probability of a collision when the first retry is attempted.

Ethernet is the classic CSMA/CD protocol. Non-switched Ethernet uses a broadcast delivery mechanism in which each frame that is transmitted is heard by every station. CSMA/CD is an arbitrary access method that provides a method to detect and recover from simultaneous transmissions. Each interface monitors the network for a carrier signal (Carrier Sense). During a gap between transmissions, each interface has an equal chance to transmit data (Multiple Access). If two interfaces try to transmit data at the same time, the transceiver circuitry detects a transmit collision (Collision Detection). Both interfaces must wait a short period of time before they attempt to resend data. The wait period is determined by using an exponential back-off algorithm.

Algorithms of CSMA/CD accesses the network was originally developed for the original Ethernet topology. Ethernet originally consisted of a single-wire, bidirectional backbone. The theory of operation is still the same today, but Ethernet topologies use more advanced components that allow a higher transmission rate.

1. The host has a message
2. Is there traffic on the network. If yes go to step 2 (loop until no traffic )
3. The host sends the message
4. Was there a collision ? If no -- success go to step 7.
5. Send the jam signal
6. Wait. Back off exponentially. Then go to step 2 (collision loop)
7. Success

See also the similar Carrier sense multiple access with collision avoidance (CSMA/CA) protocol.

Half-duplex and Full Duplex Mode

Duplex systems are employed in nearly all communications networks, either to allow for a communication "two-way street" between two connected parties or to provide a "reverse path" for the monitoring and remote adjustment of equipment in the field.

• A half-duplex system provides for communication in both directions, but only one direction at a time (not simultaneously). Typically, once a party begins receiving a signal, it must wait for the transmitter to stop transmitting, before replying.

  An example of a half-duplex system is a two-party system such as a "walkie-talkie" style two-way radio, wherein one must use "Over" or another procedure to indicate the end of transmission, and ensure that only one party transmits at a time, because both parties transmit on the same frequency. A good analogy for a half-duplex system would be a one lane road with traffic controllers at each end. Traffic can flow in both directions, but only one direction at a time with this being regulated by the controllers.
   

• A full-duplex system allows communication in both directions, and unlike half-duplex, allows this to happen simultaneously. Land-line telephone networks are full-duplex since they allow both callers to speak and be heard at the same time. A good analogy for a full-duplex system would be a two lane road with one lane for each direction. Examples: Telephone, Mobile Phone, etc.

  Two way radios can be, for instance, designed as full-duplex systems, which transmit on one frequency and receive on a different frequency. This is also called frequency-division duplex. Frequency-division-duplex systems can be extended to farther distances using pairs of simple repeater stations, owing to the fact the communications transmitted on any one frequency always travels in the same direction.

Ethernet Statistics

netstat [-a] [-i] [ -e] [-n] [-o] [-p Protocol] [-r] [-s] [Interval]

Where:

• - i  : Displays statistics (Solaris). Use -e with Microsoft implementation of netstat. This parameter can be combined with -s.
• -a : Displays all active TCP connections and the TCP and UDP ports on which the computer is listening.
• -n : Displays active TCP connectionsbut without DNS resolution of addresses
• -o : Displays active TCP connections and includes the process ID (PID) for each connection. You can find the application based on the PID on the Processes tab in Windows Task Manager. This parameter can be combined with -a, -n, and -p.
• -p Protocol : Shows connections for the protocol specified by Protocol. In this case, the Protocol can be tcp, udp, tcpv6, or udpv6. If this parameter is used with -s to display statistics by protocol, Protocol can be tcp, udp, icmp, ip, tcpv6, udpv6, icmpv6, or ipv6.
• -s : Displays statistics by protocol. By default, statistics are shown for the TCP, UDP, ICMP, and IP protocols. If the IPv6 protocol for Windows XP is installed, statistics are shown for the TCP over IPv6, UDP over IPv6, ICMPv6, and IPv6 protocols. The -p parameter can be used to specify a set of protocols.
• -r : Displays the contents of the IP routing table.

Examples:

• To display both the Ethernet statistics , type the following command: netstat -i # Solaris
• To display the statistics for only the TCP and UDP protocols, type the following command: netstat -s -p tcp udp
• To display active TCP connections and the process IDs every 5 seconds, type the following command: netstat -o 5
• To display active TCP connections and the process IDs using numerical form, type the following command: netstat -n -o

Collision rates

A collision is the mechanism used by Ethernet to control access and allocate shared bandwidth among stations that want to transmit at the same time on a shared medium. Because the medium is shared, a mechanism must exist where two stations can detect that they want to transmit at the same time. This mechanism is collision detection.

Collision rates higher then 5% on 10-Mbps LAN and 10% on 100-Mbps LAN are first indications of the network overload.

Switches minimize collisions so this is not a problem in modern LAN.

Input/Output Errors

If netstat reports significant number of i/o errors this is usually indication of duplicate IP address on the network. Other possible reasons include faulty cable of RJ45 jack, a faulty port of the network switch or a faulty LAN card in the computer.

Ethernet Addresses

This material is partially copied from SA-300-S10 Sun Solaris network administration manual

An Ethernet address is the device’s unique hardware address. An Ethernet address is sometimes referred to as a media access control (MAC) address. An Ethernet address is 48 bits long and is displayed as 12 hexadecimal digits (six groups of two digits) separated by colons. An example of an Ethernet address is  44-45-53-54-42-00

• The IEEE administers unique Ethernet addresses. IEEE designates the first three octets as vendor-specific. Most Sun systems begin with the sequence 08:00:20. The Sun Enterprise 10000 and Sun Fire 15K systems begin with 00:00:be, and the SunBlade systems begin with 00:03:ba. Sun assigns the last three octets to the products it manufactures to ensure that each node on an Ethernet network has a unique Ethernet address.

The IEEE specification enables the vendor to decide whether to use the host-based addressing approach or the port-based addressing approach. By default, Sun uses host-based addressing on its networks interface cards (NICs).

The network interface drivers in Sun systems obtain the Ethernet address for the Ethernet interface from a system’s hardware. For example, desktop systems use the address in the nonvolatile random access memory (NVRAM) chip, while some large server systems obtain their address from a special board installed in the system. By default, all interface addresses on a system use just one Ethernet address, either the NVRAM or the special board, even though each Ethernet interface controller has a built-in Ethernet address.

For systems configured to have more than one interface on the same physical subnet, you need a unique Ethernet address that is different from the primary host-based assigned Ethernet address.

There are three types of addresses: unicast, broadcast, and multicast.

• Unicast Addresses. Unicast addresses are used for one-to-one communication. The system uses a unicast address to send a message to another system on the local Ethernet network. You can use a system’s unique Ethernet address as a unicast address.
• Broadcast Addresses. A device uses a broadcast address to send messages to all systems on the local Ethernet network. The Ethernet broadcast address is represented in the form of all 1s in binary format and as ff:ff:ff:ff:ff:ff in hexadecimal format. When the Network Interface layer receives an Ethernet frame with a destination address of all 1s, it passes the address to the next layer for processing.
• Multicast Addresses. A system uses a multicast address to send a message to a subset of systems on the local Ethernet. In Ethernet multicast addressing, the value of the first three octets determines if the address is multicast. The last three octets determine the specific multicast’s group identity.

Setting a Local Ethernet Address

In today’s network environments, many systems have multiple interfaces, often on the same subnet or collision domain. Because an Ethernet address targets systems, each interface on the same network or subnet on a multi-interface system must have a unique Ethernet address. Sun network adapters have local Ethernet addresses encoded in their programmable read-only memories (PROMs).

To view the current host-based Ethernet address, perform the command at the ok prompt:

ok banner
Sun Ultra 5/10 UPA/PCI (UltraSPARC-IIi 360MHz), No Keyboard
OpenBoot 3.19, 128 MB (50 ns) memory installed, Serial #12153379.
Ethernet address 8:0:20:b9:72:23, Host ID: 80b97223.
ok

To display the Ethernet address assigned to each interface, perform the command:

# ifconfig -a

lo0: flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 1
inet 127.0.0.1 netmask ff000000

hme0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
inet 192.168.30.31 netmask ffffff00 broadcast 192.168.30.255
ether 8:0:20:b9:72:23

qfe0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 3
inet 192.168.1.1 netmask ffffff00 broadcast 192.168.1.255
ether 8:0:20:b9:72:23

Set the local-mac-address? variable in the system’s electrically erasable programmable read-only memory (EEPROM) to enable the use of port-based Ethernet addresses.

To view the contents of the EEPROM for the definition of the local-mac-address? variable, perform the command:

# eeprom local-mac-address?
local-mac-address?=false

You can set the local MAC address to true, which enables network drivers to use their own port-based addresses after reboot and not the system default host-based addressing by performing the command:

# eeprom local-mac-address?=true

The ifconfig ether command can also configure port-based addressing. This might be necessary if the interface card cannot supply its own unique Ethernet address. You can change the interface Ethernet address of 8:0:20:f0:ac:61 from a globally assigned Ethernet address to a locally assigned address of 0a:0:20:f0:ac:61 by changing the seventh bit to 1, and assigning a local unique number to the last 3 bytes.

To change the Ethernet address, perform the command:

# ifconfig hme1 ether 0a:0:20:f0:ac:61

To verify a change in the Ethernet address, perform the command:

# ifconfig hme1
hme1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
inet 192.168.30.31 netmask ffffff00 broadcast 192.168.30.255
ether a:0:20:f0:ac:61

This change of the Ethernet address is effective until you reboot the system. To make the change permanent, modify the /etc/rc2.d/S72inetsvc script by using the ifconfig command with the correct Ethernet address.

Ethernet frame format

This material copied from IBM Redbook

Know how to interpret the data contained in your Ethernet frames by understanding the frame formatting.

While tracing LAN communications, you might need to look at the transmitted frames. To understand the data that is contained in the frame, you must know how it is formatted. The figures below show the frame format of two Ethernet standards: IEEE 802.3 and Ethernet version 2.

The Frame Check Sequence (FCS) is a part of the frame put in place to verify that the information each frame contains is not damaged during transmission. If a frame is corrupted during transmission, the FCS on the frame will not match with the recipient's calculated FCS. Any frames that do not match the calculated FCS will be discarded.

IEEE 802.3 frame format

Ethernet version 2 frame format

Ethernet version 2 supports SNA by placing the IEEE 802.2 LLC header and data into the information field. It also puts value X'80D5' into the Type field. This shows the frame format.

Maximum Transfer Units

The maximum transfer unit (MTU) is the largest amount of data that can be transferred across a physical network. The Ethernet MTU is hardware  specific. For a physical Ethernet interface, the MTU is 1500 bytes, while the MTU is 8232 bytes for a loopback interface. The loopback interface is a pseudo device that communicate or loops back to the host itself.

Ethernet Frame Errors

Ethernet frames can be significantly damaged when they traverse a network. When a host receives a frame, the Ethernet interface performs integrity checking to verify Ethernet frame validity. Some of these error conditions are:

• Runts Packets that are less than 64 bytes are too short and are discarded. Runts are usually caused by collisions. These can be formed by poor wiring and electrical interference.
• Jabbers Packets that are greater than 1500 bytes (MTU) are too long and are discarded. These indicate that a device has electrical problems.
• Long A frame that is between 1518 and 6000 bytes long, often because of faulty hardware or software on the sending system.
• Giant A frame that is more than 6000 bytes long, often because of faulty hardware or software on the sending system.
• Bad CRC If the received packet fails the CRC, the packet is corrupted and discarded. This is also known as a frame check sequence (FCS) error.

Our scenario assumes a second NIC is being installed into a previously configured networked system. Before installing the new hardware, save the current system configuration by running prtconf -vD. The output of this command can be extensive on larger systems, so it's best to direct the output of prtconf to a file.

After physically installing the new hardware, boot the system with the -r option from the OK prompt. This will have the system scan for new hardware and build the device driver directories accordingly. When that has been completed, run prtconf -vD again and compare its content to the previous execution's results. If a new device doesn't show up, consult the instruction manual for that device as it may require a new driver to be installed or some other special action specific to that hardware.

The name of a NIC is the driver abbreviation with the instance number (a consecutive number starting from 0 counting each device that uses that driver). If you cannot tell which driver runs your hardware, consult the NIC card's manual.

The results of an ifconfig -a show us the current state of the network devices:

lo0: flags=1000849 mtu 8232 index 1
        inet 127.0.0.1 netmask ff000000 
hme0: flags=1000843 mtu 1500 index 2
        inet 192.168.1.100 netmask ffffff00 broadcast 192.168.1.255
        ether 

The built in NIC (instance #0) had previously been configured, and lo0 is an instance of the local loopback address. Since hme1 doesn't show up, it means it has yet to be initialized and configured.

IP Configuration

Persistent IPv4 Configuration

In order to have the system configure our NIC at boot, the first step is to get an IP address and subnet mask. In our case, we are going to put this second NIC into a different IP range than the original NIC. Our first NIC is in the 192.168.1.x network, so we will put the new NIC into the 192.168.2.x network. Both of these networks have a subnet mask of 255.255.255.0. Note: Always make sure the assigned IP is not already taken up by another machine; to do so, use ping from a machine already configured in that network.

Next we add a line to the /etc/hosts file for our new card:

192.168.2.100		host2.mydomain.com		host2

Now, we create a file in /etc that is named hostname. For example, our first NIC's file is /etc/hostname.hme0. Our new device, hme1, will need the file /etc/hostname.hme1. In this file, we will put the name associated with the IP (as found in the /etc/hosts file). It should be the first name in the /etc/hosts file. In our scenario, /etc/hostname.hme1 should contain:

host2.mydomain.com

Then we edit the /etc/netmasks file for our new network:

192.168.2.0	255.255.255.0

Reboot the system, and your network card has been configured for the new network with the proper subnet mask. You can check it by running an ifconfig -a again:

lo0: flags=1000849 mtu 8232 index 1
        inet 127.0.0.1 netmask ff000000 
hme0: flags=1000843 mtu 1500 index 2
        inet 192.168.1.100 netmask ffffff00 broadcast 192.168.1.255
        ether 
hme1: flags=1000843 mtu 1500 index 3
        inet 192.168.2.100 netmask ffffff00 broadcast 192.168.2.255
        ether 

Non-Persistence IPv4 Configuration

To configure the NIC without having to reboot (having previously installed the hardware), you first need to initialize or plumb the network card:

	ifconfig hme1 plumb

Then you configure the device:

	ifconfig hme1 192.168.2.100 netmask 255.255.255.0

Now, you just bring the NIC online.

	ifconfig hme1 up

Your network card is now up and running. You will still need to make the preceding file modifications or your card will not be configured upon reboot.

IPv6 Configuration

Almost everything with IPv6 is designed to be automatic. All that is needed is to tell the system that IPv6 is to be used, and it will handle the rest. To configure it at boot (persistent), execute the following command:

	touch /etc/hostname6.interface 

To enable IPv6 at the command line (which will be lost when the system reboots), run the following commands:

	ifconfig  inet6 plumb
	ifconfig  inet6 up

Troubleshooting

The following is a collection of tools that might be useful for troubleshooting of  networking problems:

 /bin/netstat: Many netstat options are useful in troubleshooting. Some examples

• netstat -i # General information with statistics on each NIC. Often used in homemade monitoring scripts.
• netstat -a. # Port/Socket information can be had with a

   This is a good way to tell what remote systems are attaching to what ports, in case you have network daemons that are locking up or if you are wondering what computers are browsing your web server.

• netstat -rn # most popular way to check current system routing tables with .

  The n option should almost always be used when troubleshooting network issues, since DNS lookups for IP-to-Hostname lookups are not local services. The entry to double check is your "default" destination, since this is the system's default router or gateway.

/usr/sbin/snoop: Root can run snoop to capture network traffic and display packet content on the screen. Snoop is useful for finding malfunctioning NICs flooding the network or for ensuring that DHCP is working properly. Before you configure a NIC, you can run snoop to verify that you are seeing packets flowing through your NIC that is, the cable/NIC/switch port are working), and the IP addresses of those packets are consistent with what is expected for the NIC's new network. Some examples:

• snoop -d hme0 broadcast #only broadcast frames in summary mode
• snoop -d qfe0 -V 192.168.1.2
• snoop -d qfe0 -o /tmp/snooper 192.168.1.2 # capture to a file
• snoop -d qfe0 -o /tmp/snooper broadcast # same with broadcast
• snoop -i /tmp/snooper -V | egrep -iv 'nfs|ack|contin|ftp|ip'

/sbin/ifconfig:  An ifconfig -a will give you information about all NIC, including IP address (inet), subnet mask (netmask), and if run as root, the MAC address (ether). Please note that only root gets full information from this command.

/usr/sbin/arp: arp allows you to view that table to see what kind of information your computer has cached, as well as to input or delete entries, in case of problems. When having DNS issues, an arp command with the -an option will list the entire table without resolving IP address to host names. arp is one way to get the MAC address of the NIC answering for a host.

Recommended Links

• CISCO Ethernet page
• Ethernet - Wikipedia, the free encyclopedia
• Charles Spurgeon's Ethernet (IEEE 802.3) Web Site Charles is the author of O'Reilly book on the subject.
  • Table of Contents
  • Chapter 13: Multi-Segment Configuration Guidelines  - online chapter 
  • Errata: List of errors found in Ethernet: The Definitive Guide
  • Other Books on Ethernet
• Howstuffworks How Ethernet Works
• Ethernet
• Ethernet [Charles Spurgeon]  - excellent info on 10 and 100-Mbps Ethernet (IEEE 802.3)
• PC Lube and Tune Ethernet Page useful document dated 12 Apr 1995
• Ether BLAM - Binary Logarithmic Arbitration - an enhanced backoff algorithm that is backwards compatible with existing CSMA/CD Ethernet (802.3w).
• Ethernet [H. Gilbert] - the most common choice for high speed communication to the desktop
• Networks at Home - how to install an local area network.
• Troubleshooting Numbers for Ethernet - maintained by Charles Spurgeon.

FAQs

• FAQ - comp.dcom.cabling
• FAQ - comp.dcom.lans.ethernet

Reference

Ethernet Codes - Michael A Patton's list of IEEE OUIs

OUI and Company ID Assignments - used to generate Universal LAN MAC addresses.

Quiz

Q1. Fast Ethernet : Max distance for UTP and Fiber ?

a. 200m, 1000m

b. 1000m, 2000m

c. 100m, 1000m

d. 100m, 2000m

A:  D

Q2. How long is MAC address (bytes) ? How may bytes identifies the vendor?

a. 8,2

b. 6,3

c. 8,4

d. 6,2 

A: B

Q3. What is a "preamble" in IEEE 802.3 frame?

a. an indication of the end of a frame

b. an indication of the start of a new frame

c. an indication of congestion

d. an indication of a corrupted frame

A: B

Q4. What does a "preamble" consist of ?

a. all 0s

b. all 1s

c. a 1 in the beginning and rest 0s

d. alternating 0s and 1s

A: D

Q5.  The two most popular carrier sensing schemes are Token passing and ""_______""".

A: CSMA/CD  (Carrier-Sense Multiple Access with Collision Detection)

Q6. This type of access method is generally used by Ethernet to regulate network traffic on a main cable segment.

A: CSMA/CD

Q7. Which network conforms to IEEE 802.3 specs and is currently is considered to be "standard Ethernet" ?

A: 100BaseTX

Q8: A frame greater than the MTU is known as:

A: c  

Q9. ""______" is a device which  provide communication between the computer and the LAN cable and is directly connected to the vampire tap to which the cable is attached.

A: Transceiver

Last modified: February 28, 2008