Ethernet Technical Overview

 <  Free Open Study  >  

Ethernet is extremely popular, and many fine books and whitepapers have been written on it. For these reasons, we assume that you have some background in Ethernet technology; therefore, this chapter does not cover Ethernet frames, hubs, and cables in any great length. You should become familiar with the different Ethernet frame types, DIX Version II, and 802.2 frames , as well as the different media types used in Ethernet. This chapter instead focuses more on Spanning Tree, Fast Ethernet, Gigabit Ethernet, and Ethernet and Token Ring switching.

Ethernet Operation

Ethernet operates at the OSI Layer 2, the data link layer. The data link layer actually is divided into two sublayers : the MAC layer and the Logical Link Control (LLC) layer. The LLC layer ”802.2, in this case ”is a standardized interface between a hardware-specific MAC and a Layer 3 protocol.

The MAC layer has the following responsibilities:

  • Generating the physical source and destination address for a frame. These are 48-bit industry-wide unique addresses; the first 3 bytes are assigned by the IEEE, and the last 3 bytes are vendor-unique.

  • Ensuring reliable transmissions.

  • Synchronizing data transmissions.

  • Performing error recognition.

  • Performing flow control.

Table 2-2 lists the common physical characteristics of 10-Mbps, 10-Mbps, and 1000-Mbps Ethernet.

Table 2-2. Common Ethernet Specifications
Specification 10-Mbps 100-Mbps 1000-Mbps
Minimum frame 512 bits/64 bytes 512 bits/64 bytes 4096 bits/512 bytes
Bit time ( graphics/u0956.gif s) 0.1 graphics/u0956.gif s 0.01 graphics/u0956.gif s 0.001 graphics/u0956.gif s
Maximum round-trip delay ( graphics/u0956.gif s) 51.2 graphics/u0956.gif s 5.12 graphics/u0956.gif s 4.096 graphics/u0956.gif s
Maximum network diameter, with no repeaters ( meters ) Approximately 45710 milleseconds 457 milleseconds 3661 milleseconds
Maximum number of repeaters in a collision domain Approximately 5 1 Class I repeater or 2 Class II repeaters 1
Ethernet CSMA/CD

Ethernet technology commonly is referred to as carrier-sense multiple access collision detect (CSMA/CD). Ethernet transmits frames in the following manner:

  1. Carrier sense ” This also is known as "listen before talking." An Ethernet station wanting to transmit a frame listens to the medium before it transmits to ensure that the medium is available.

  2. Talk if quiet ” If the channel is quiet for a specific amount of time, the interframe gap (IFG) before the station might begin a transmission. If the channel is busy, it is monitored until it becomes free for the length of IFG timer; after that time, transmission might begin.

  3. Collision ” A collision is measured as an excess of voltage on the "cable" or medium. A collision usually is caused by two stations transmitting data at the same time. If a collision occurs, both frames are destroyed .

  4. Collision detection ” If a station detects a collision during transmission, that transmission immediately stops. A signal jam also is sent on the media to destroy any fragmented frames, preventing corrupted data.

  5. Backoff ” After a collision, a stations waits a period of time called the backoff period. The backoff timer is a random timer generated by a backoff algorithm. This prevents all stations from trying to transmit at the exact same time after a collision. After the backoff timer expires , the station attempts to retransmit the frame. If another collision happens, the station keeps trying to retransmit the frame for 16 times. After 16 unsuccessful attempts, the frame is dropped.

Half- and Full-Duplex Ethernet

Ethernet was developed on older coax type cables capable of transmitting or receiving a signal at any given time. This is precisely why Ethernet needed CSMA/CD technology. With the advent of switches, running Ethernet over UTP and fiber, full-duplex Ethernet became available. Full-duplex Ethernet allows a station to simultaneously transmit and receive data. Ethernet frames are transmitted and received simultaneously on two pairs of UTP or fiber at any given time. Full-duplex Ethernet is essentially Ethernet without CSMA/CD. Full-duplex mode basically doubles the bandwidth of Ethernet! To run full-duplex Ethernet, the station and switch both must be capable and configured for full-duplex operation. A hub with multiple stations cannot operate in full-duplex mode.

NOTE

A station not operating in the correct duplex mode will generate an enormous number of collisions on the port where it is located. These collisions mostly likely will be registered as "late collisions." Be sure that the port on the switch and the end station are operating in the same duplex mode.


Fast Ethernet

In May 1995, the IEEE adopted the Fast Ethernet standard, 802.3u. Years later, after battling FDDI, 100VG AnyLAN, and ATM, this standard became the prominent type of LAN. As costs per port dropped along with the prices of network interface cards (NICs), Fast Ethernet won out over FDDI and 100VG AnyLAN and ATM for many reasons:

  • It allows a clean and inexpensive migration path from existing 10-Mbps Ethernet networks. At first, it could run only fiber and Cat 5 UTP, but as Table 2-1 indicates, it now can run over almost every media type.

  • Fast Ethernet didn't require costly fiber connections and didn't require complex configurations.

  • Fast Ethernet addresses QoS with enormous amounts of bandwidth, while relying on the upper layers or network design to provide QoS.

  • Basically, Fast Ethernet became a plug-and-play tenfold upgrade of the LAN. 100-Mb networks evolved out of 10-Mb networks across data centers everywhere.

Some important features and specifications of Fast Ethernet are as follows :

  • The 100Base MAC uses the original Ethernet MAC operating at 10 times the speed. This is completely backward compatible with 10-Mbps Ethernet.

  • 100Base-T includes a specification for an MII interface. A MII interface is a 100-Mbps version of the AUI adapter.

  • Fast Ethernet supports full- and half-duplex functionality.

  • Fast Ethernet operates over a wide array of different physical layers: Cat 5, Cat 3, fiber, and so on, as listed in Table 1-2.

Gigabit Ethernet

The ink was hardly dry on the Fast Ethernet standard, and the IEEE started to work on the 802.3z, or Gigabit Ethernet, standard. Three short years later, in June 1998, the 802.3z standard officially was adopted. For the most part, the Gigabit standard is the Fast Ethernet standard multiplied by 10. This is why 10-Gb Ethernet products are right around the corner and 100-Gb Ethernet is an eventuality.

Some important features and specifications of Gigabit Ethernet are as follows:

  • Gigabit Ethernet uses the 802.3 frame format, identical to that of 10-Mbps and 100-Mbps Ethernet.

  • It includes a specification for a Gigabit MII (GMII). Unlike 10-Mbps Ethernet and 100-Mbps Ethernet, the GMII is an electrical specification and does not include a physical connector. Cisco's physical Gigabit interfaces are called GBICs. The type of GBIC determines the physical gigabit connection. There are currently multimode fiber (MMF), single-mode fiber (SMF), and UTP GBICs, as well as a Cisco proprietary GBIC called a Gigastack.

One element that makes Gigabit Ethernet one of the most versatile protocols in years is the concept of the GMII. Except for the strictly 1000Base-TX switches, most Gigabit Ethernet switches come with open ports for GBICs. Depending on your network needs, you can put any type of GBIC in this port. With the click of a GBIC, your network can change from a limited 100-m copper -based network to a 10,000-m fiber-based network! The sections that follow cover the more common GBICs, Gigabit standards, and length limitations.

1000Base-SX Gigabit Ethernet

1000Base-SX GBICs use a laser-based wavelength of 850 nms. Depending on the cable type, SX GBIC operates at distances of 220 m to 550 m, as listed in Table 2-3. Wavelengths up to 850 nm are visible to the human eye.

Table 2-3. 1000Base-SX Cable Limitations
Standard Cable Size (Microns) Max. Distance (m)
1000Base-SX 62.5-um multimode fiber 275 m
1000Base-SX 50-um multimode fiber 550 m
1000 Base-LX Gigabit Ethernet

LX GBICs use a laser-based wavelength of 1300 nms. Depending on the cable type, LX GBIC operate at distances of 550 m to 5000 m, as listed in Table 2-4. Cisco also supports an LH and LX GBIC, which extends the IEEE 1000Base-LX maximum distance of 5 km.

Table 2-4. 1000Base-LX Cable Limitations
Standard Cable Size (Microns) Max. Distance (m)
1000Base-LX 62.5-um multimode fiber 550 m
1000Base-LX 50-um multimode fiber 550 m
1000Base-LX 9/10-um single-mode fiber 5 km
1000Base-LH 62.5-um multimode fiber 550 m
1000Base-LH 50-um multimode fiber 550 m
1000Base-LH 9/10-um single-mode fiber 10 km
1000Base-ZX 9/10-um single mode fiber 70 km
1000Base-ZX 9/10 Disposition-shifted fiber 100 km
1000Base-CX Gigabit Ethernet

The CX standard is for operating Gigabit Ethernet over copper for short distances. 1000Base-CX uses a 150-ohm balanced shielded copper cable. The distance of the CX standard is limited to only 25 meters.

1000Base-T Gigabit Ethernet

The IEEE standard for Gigabit Ethernet transmission over Cat 5 UTP is 802.3ab. The standard defines the maximum distance to be 100 meters and the copper to be at least Cat 5 using four pairs of wires, terminated with an RJ-45 jack. Figure 2-1 illustrates a GBIC.

Figure 2-1. GBIC Installation

graphics/02fig01.gif

NOTE

The Cisco Gigastack GBIC is a Cisco proprietary GBIC used on an uplink port to connect Gigabit Ethernet switches.


Ethernet Autonegotiation

To simplify the configuration of Ethernet devices, the 802.3.u committee defined Fast Link Pulse (FLP). FLP sends a series of pulses on the network that can deduce what duplex and speed the link is operating at. The station and the hub/switch agree on the highest priority and configure the station in that manner. Table 2-5 lists the priority that FLP uses. Both devices need to support autonegotiation logic for autonegotiation to work.

Table 2-5. Ethernet Autonegotiation Prioritization
Priority Speed and Duplex Settings
1 [*] 100Base-T2 full duplex
2 100Base-T2 half duplex
3 100Base-TX full duplex
4 100Base-T4 half duplex
5 100Base-TX half duplex
6 10Base-T full duplex
7 10Base-T half duplex

[*] The priority of 1 eventually might be replaced by 1000Base-T full-duplex, shifting the others down by one.

Infrastructure devices, such as routers and servers, always should have speed and duplex settings fixed. Most 100-Mbps and greater NICs support full-duplex operation. Running at full-duplex mode essentially doubles the capacity of Ethernet. Taking advantage of this is the cheapest network upgrade that you will ever do!

NOTE

Duplex modes are a function of the hardware built into the network interface card (NIC). Software upgrades will not allow you to run full-duplex mode. For full-duplex mode to work, the station and the switch port must be capable of full-duplex operation.


 <  Free Open Study  >  


CCIE Practical Studies, Volume I
CCIE Practical Studies, Volume I
ISBN: 1587200023
EAN: 2147483647
Year: 2001
Pages: 283
Authors: Karl Solie

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net