3 Use the data link layer of the broadcast channel

Broadcast channels can carry one-to-many communications . LAN is the broadcast channel used .

3.1 LAN

LAN refers to a group of computers interconnected by multiple computers in a certain area , It's usually within a few thousand meters ; Wan usually spans a large physical range , It covers tens to thousands of kilometers , It can connect multiple cities or countries , Or across several continents and providing long-distance communication , Forming an international remote network .

Protocol running on the cables connected to the devices at both ends , For LAN , There's Ethernet , Token ring ,FDDI etc. .

The following figure shows the topology of LAN :

3.1.1 The classification of LAN

• Ethernet ： Ethernet is the most widely used LAN , Including standard Ethernet （10Mbps）、 Fast Ethernet （100Mbps）、 Gigabit Ethernet （1000 Mbps） and 10G Ethernet , They all fit IEEE802.3 Series of standard specifications . Logical topology master line , Physical topology is star or extended star . Use CSMA/CD.
• Token ring network ： Physically, star topology is adopted , Logically, it is a ring topology . Already “ chrysanthemums after the double ninth festival ”
• PDDI network （ Floer Distrilouked Data inervece）： Physically, the double ring topology is adopted , Logically, it is a ring topology .
• ATM network （ Asynchronous Transler Mode）： Newer switching unit technology , Use 53 Byte fixed length units are exchanged .
• WLAN ( Local Area Network WLAN)： use IEEE 802.11 standard .

3.1.2 IEEE 802 standard

IEEE 802 The series standard is IEEE 802 LAN/MAN LAN developed by the Standards Committee 、 Technical standard of man （1980 year 2 Month set up ）. One of the most widely used is Ethernet 、 Token ring 、 WLAN, etc . Each sub standard in this series of standards is the responsibility of a special working group in the Committee .

• IEEE 802.3: Ethernet media access control protocol (CSMACD） And physical layer technical specifications 2.
• IEEE 802.5: Token ring network (Token-Ring) Media access control protocol and physical layer technical specification .
• IEEE 802.8: Optical fiber technical advisory group , Provide technical consultation on optical fiber networking .
• IEEE 802.11: WLAN (WLAN) Media access control protocol and physical layer technical specification .

3.1.3 MAC Sublayer LLC Sublayer

IEEE 802 The LAN reference model described in the standard only corresponds to OSI The data link layer and physical layer of the reference model , It divides the data link layer into logical link layer LLC Sublayer and media access control MAC Sublayer .

• Logical link control LLC（Logical Link Control） - Deal with content unrelated to access to the transmission media
• Media access control MAC（Medium Access Control）- Deal with content related to access to transmission media

However, due to the rapid development of the Internet TCP/IP The only Lan that the system often uses is DIX Ethernet V2 instead of IEEE 802.3 LAN in standard , So now IEEE The logical link control sublayer designated by the committee LLC The role of has disappeared , Adapters produced by many manufacturers are only equipped with MAC Agreement without LLC agreement

[ Failed to transfer the external chain picture , The origin station may have anti-theft chain mechanism , It is suggested to save the pictures and upload them directly (img-uVJ74rlZ-1651218479207)(https://note.youdao.com/yws/res/19519/WEBRESOURCE0aa6f4aa09dfcf6e9af989fbb48aac11)]

3.2 Ethernet

Ethernet (Ethernet) Refers to the Xerox The company was founded by Xerox、Intel and DEC Baseband bus LAN specification jointly developed by the company , It is the most common communication protocol standard adopted by the existing LAN . Ethernet use CSMA/CD（ Carrier monitor multiple access and collision detection ） technology .

Ethernet plays a dominant role in all kinds of LAN technologies ：

1. Low cost （ Ethernet network card is missing 100 block ）;
2. It is the most widely used LAN technology ;
3. Than token ring 、ATM Net cheap , Simple ;
4. Meet the network speed requirements ：10Mb/s~10Gb/s.

Ethernet has two standards ：

1. DIX Ethernet V2： First LAN product （ Ethernet ） Statute .
2. IEEE 802.3：IEEE 802 The committee 802.3 The first one developed by the working group IEEE Ethernet standard for .（ There are a few changes to the frame format ）

Ethernet provides no connection 、 Unreliable service .

3.2.1 The frame format

So far, TCP/IP The LAN often used in the system refers to DIX Ethernet V2, Most of these data frames use Ethernet II Format .

The difference between the two types of frame formats ：

3.2.1.1 Ethernet_II The frame format

When Type The field value is greater than or equal to 1536 （ Or hexadecimal 0x0600） when , The frame uses Ethernet II Format .
Most data frames in Ethernet use Ethernet II Format .

The length of Ethernet data frame is 64-1518 Between bytes .
Ethernet_II The fields in the frame are described as follows ：

• DMAC（Destination MAC） It is purpose MAC Address .DMAC The length of the field is 6 Bytes , Identify the receiver of the frame .
• SMAC（Source MAC） It is the source. MAC Address .SMAC The length of the field is 6 Bytes , Identify the sender of the frame .
• The type field （Type） Used to identify the high-level protocol contained in the data field , The length of this field is 2 Bytes . The value of type field is 0x0800 The frame of the image represents IP Protocol frame ; The value of type field is 0806 The frame of the image represents ARP Protocol frame .
• Data field (Data) It's network layer data , The minimum length must be 46 Bytes to ensure that the frame length is at least 64 byte , The maximum length of a data field is 1500 byte .
  Cyclic redundancy check fields （FCS） Provides an error detection mechanism . The length of this field is 4 Bytes .

3.2.1.2 IEEE802.3 The frame format

When Type The field value is less than or equal to 1500（ Or hexadecimal 0x05DC） when , The frame uses IEEE 802.3 Format .

IEEE 802.3 The frame length field value is less than or equal to 1500 (0x05DC)

IEEE 802.3 The frame format is similar to Ethernet_II frame , It's just Ethernet_II The frame of Type Domain by 802.3 The frame of Length Domain substitution , And occupied Data Field 8 Bytes as LLC and SNAP Field .

• Length The field defines Data The number of bytes the field contains .
• Logical link control LLC（Logical Link Control） By the destination service access point DSAP（Destination Service Access Point）、 Source service access point SSAP（Source Service Access Point） and Control Field composition .
• SNAP（Sub-network Access Protocol） By organization code （Org Code） And type （Type） Field composition .Org code All three bytes are 0.Type The meaning of the field is the same as Ethernet_II In the frame Type The fields are the same .

IEEE802.3 Frame according to DSAP and SSAP The values of fields can be divided into the following categories ：

• When DSAP and SSAP Take a specific value 0xff when ,802.3 The frame becomes Netware-ETHERNET frame , Used to carry NetWare Data of type .
• When DSAP and SSAP Take a specific value 0xaa when ,802.3 The frame becomes ETHERNET_SNAP frame .ETHERNET_SNAP Frames can be used to transmit a variety of protocols .
• DSAP and SSAP Other values are pure IEEE802.3 frame .

3.2.2 Data frame transmission

The data link layer is based on MAC Address for frame transmission

Ethernet passes through the layer-2 link MAC Address to uniquely identify network devices , And realize the communication between network devices on LAN .MAC Address is also called physical address , Most network card manufacturers put MAC The address is burned into the network card ROM in . The sender uses the of the receiver MAC Address as destination address . After the Ethernet frame is encapsulated, it will be converted into a bit stream through the physical layer and transmitted on the physical medium .

in addition , It should be noted that network packets are transmitted in large end byte order , Of course, the header has to be big endian .Java The default is also big end processing .

3.2.3 The frame type

Ethernet depends on your destination MAC Can be divided into ： unicast , radio broadcast , Multicast .

3.2.3.1 unicast

Sending from a single source to a single destination . Each host interface consists of a MAC Address unique identification ,MAC Address of the OUI in , The first byte is 8 The bits represent the address type . For hosts MAC Address , This bit is fixed to 0, Show purpose MAC The address is MAC Address frames are sent to a unique destination .

In the conflict domain , All hosts can receive unicast frames sent by the source host , But other hosts find that the destination address is different from the local address MAC If the address is inconsistent, the received frame will be discarded , Only the real destination host will receive and process the received frame .

The following figure shows the unicast data frame

3.2.3.2 radio broadcast

Indicates that frames are sent from a single source to all hosts on the shared Ethernet . The purpose of the broadcast frame MAC The address is hexadecimal FF:FF:FF:FF:FF:FF, All hosts that receive the broadcast frame will receive and process the frame .

Broadcast mode will generate a lot of traffic , Leading to lower bandwidth utilization , And then affect the performance of the whole network .

When all hosts in the network need to receive the same information and process it , It's usually broadcast .

The following figure shows the broadcast data frame

3.2.3.3 Multicast

Multicast is more efficient than broadcast . Multicast forwarding can be understood as selective broadcasting , Hosts listen for specific multicast addresses , Receive and process the purpose MAC The address is the multicast MAC The frame of the address .

Multicast MAC Address and unicast MAC The address is passed through the second... In the first byte 8 Bit by bit . Multicast MAC Address No 8 The bits are 1, And unicast MAC Address No 8 The bits are 0.

When a group of hosts on the network is needed （ Not all hosts ） Receive the same information , And other hosts are not affected , Multicast is usually used .

The following figure shows the multicast data frame

3.2.4 CSMA/CD agreement

The early development of LAN , LAN using bus mode , The feature is that all computers are on the same bus , Data can only be sent when the bus is idle . In this way, a protocol is needed to negotiate the speech conflict between multiple computers , This is it. CSMA/CD agreement ： Carrier monitoring multipoint access / Collision detection .

Because Ethernet sends data using Manchester code The signal of , Therefore, the Ethernet protocol does not need the tail of the delimiter , Because once the host no longer sends signals , The high-low level conversion in Ethernet disappears , This feature can be used as the tail of the frame .

Ethernet uses carrier frames with collision detection to listen to multiple access (CSMA/CD) Mechanism . Nodes in Ethernet can see all the information sent in the network , therefore , We say Ethernet is a broadcast network .

The working process of Ethernet is as follows , When a host in Ethernet wants to transmit data , It will proceed as follows ：

1. Whether there is a signal transmitted on the frame listening channel . If any , Indicates that the channel is busy , Just continue to listen , Until the channel is free .
2. If no frame hears any signal , Just transfer data
3. Continue to listen to frames while transmitting , If a conflict is found, the backoff algorithm is executed , Wait randomly for a period of time , Repeat the steps 1( When conflict occurs , The computer involved in the conflict will send and return to the frame listening channel state .
4. If no conflict is found, the transmission is successful , Computers all computers before trying to send data again , Must wait after the last send 9.6 Microsecond ( With 10Mbps function ).

according to CSMA/CD agreement , It can be inferred that to detect the collision in the transmission process , The minimum length of data is 64 byte . Ethernet MTU by 64~1500

In the use of CSMA/CS When the agreement , It is impossible for one station to transmit and receive at the same time （ But the channel must be monitored while sending ）. Therefore use CSMA/CS Protocol Ethernet can not carry out full duplex communication, but only two-way alternate communication （ Half duplex communication ）.

3.3 Ethernet classification

According to the different transmission rates , Ethernet is divided into standard Ethernet (10Mbit/s)、 Fast Ethernet (100Mbis) Gigabit Ethernet (1000Mbs) And 10 Gigabit Ethernet (10Gbit/s), These Ethernet are compliant with IEEE 802.3 Is compatible .

3.3.1 Standard Ethernet

Standard Ethernet is the earliest Ethernet , Its transmission rate is 10Mbts, Also known as traditional Ethernet . The networking mode of this Ethernet is very flexible , It is possible to use both rough and 、 Thin cables form a bus network , Twisted pair can also be used to form a star network , You can also use coaxial cable and twisted pair to form a hybrid network . These networks are consistent with EE8023 standard ,IEEE8023 Some traditional Ethernet physical layer standards specified in are as follows .

1. 10 Base-2： Use thin coaxial cable , The maximum segment length is 185m.
2. 10 Base-5： Use thick coaxial cable , The maximum segment length is 500m.
3. 10 Base-T： Use double lines , The maximum segment length is 100m.
4. 10 Boad-36： Use coaxial cable , The maximum segment length is 3600m.
5. 10 Base-F： Using optical fiber , The maximum segment length is 2000m, The transmission rate is 10Mb/s.

The first digit in the above standard represents the transmission rate , Unit is Mbis; The number at the end represents the length of a single segment of network cable ( The datum unit is 100m);Base Baseband transmission , Broad Indicates broadband transmission .

3.3.2 Fast Ethernet

With the development of network and the popularization of various network technologies , Standard Ethernet technology has been difficult to meet people's demand for network data traffic and rate .1993 year 10 Months ago , People can only choose expensive 、 be based on 100Mbs Fiber optic cable FDD Technology to build a high standard network ,1993 year 10 month , Grand Junction The company launched the world's first fast Ethernet hub FastSwitch10/100 And 100m network interface card Fast NIC 100, Fast Ethernet technology has been officially applied .

And then ,Intel、3COM And other companies have also launched their own fast Ethernet devices , meanwhile IEEE802 The working group is interested in 100Mbi/s Various standards of Ethernet are studied , And in 1995 year 4 Month issued IEEE 802.3u 100Base-T Fast Ethernet standard , The era of Fast Ethernet is coming .

IEEE 802.3U The standard basically maintains the provisions of the standard lan , Including frame format 、 Interface 、 Media access control method (CSMA/CD) etc. , Just change the data transfer rate from 10Mbts Promoted to 100Mbit/s, Some new physical layer standards are used , The details are as follows .

1. 100 Base-1X： Use two pairs of 5 Class shielded or unshielded twisted pair , A pair for sending data , A pair for transmitting data ; Use RJ-45 or DB9 Interface , The maximum distance between the node and the hub is 100m, Support full duplex .
2. 100 Base-T4： Use 4 Yes 3 class 、4 Class or 5 Class a twisted pair ,3 For sending data ,1 For detecting collision signals ; Use R-45 The connector , The maximum segment length is 100m, Full duplex is not supported .
3. 100 Base-FX： Use a pair of single-mode or multimode fibers , All the way for sending data , One way is used to receive data ; The maximum segment length is 200m( When using single-mode fiber 2000m), Support full duplex . This kind of network is mainly used to build the main network , To improve the transmission rate of the backbone network .

3.3.3 Gigabit Ethernet

Gigabit Ethernet (GigabitEthernet) Also known as Gigabit Ethernet .1995 year 11 month ,IEEE802.3 Any high-speed research group of the working group , Research to increase the fast Ethernet rate to 1000Mbls Feasibility and method of .1996 year 6 month ,IEEE The Standards Committee approved the application for authorization of Gigabit Ethernet scheme , And then IEEE802.3 The working group established IEEE 802.3z The Working Committee , The committee established the Gigabit Ethernet standard , The main provisions of the standard are as follows ：

• Rate is 1000Mbit/s Full duplex of Ethernet in communication / With half duplex operation .
• Use 8023 Ethernet frame format 、CSMA/CD technology .
• Support a repeater in a conflict domain .
• Backwards compatible 10Base-T and 100Base-T IEEE802.3.

The working group has developed a series of physical layer standards for Gigabit Ethernet , The common standards are as follows :

1. 1000 Base-SX： The core diameter used is 50m And 625m、 The working wavelength is 850m Multimode fiber , use 8B/10B Encoding mode , The transmission distances are 260m and 525m. This standard is mainly applied to the short distance main network on the same floor of the building .
2. 1000 Base-LX： The core diameter used is 50pm And 625m、 The working wavelength is 850nm And core diameter is 5m、 The working wavelength is 1300nm The multimode 、 Single mode fiber , The transmission distances are 525m、550m and 3000m. This standard is mainly applied to campus backbone network .
3. 1000 Base-CX： Use 1500 Shielded twisted pair , use 8B/10B Encoding mode , The transmission rate is 1.25Gbis, The transmission distance is 25m. This standard is mainly used for the connection of cluster devices , Such as equipment interconnection in a switch room .
4. 1000 Base-T： Use 4 Yes 5 Class unshielded twisted pair , use PAM5 Encoding mode , The transmission distance is 100m.

This standard is mainly used for communication of buildings on the same floor , Thus, unshielded twisted pair cable laid by standard Ethernet or fast Ethernet can be used .

Gigabit Ethernet uses optical fiber as uplink , For connection between buildings , Originally designed as a switching technology , After that, it is widely used in server connection and backbone network . Now , Gigabit Ethernet has become the mainstream network technology , Whether it's large enterprises or small and medium-sized enterprises , When building a network, Gigabit Ethernet will be the first choice for high-speed network technology .

3.3.4 10 Gigabit Ethernet

10 Gigabit Ethernet (10 Gigabit Ethernet,10GE) Also known as 10 Gigabit Ethernet , It is a high-speed Ethernet after Gigabit Ethernet . On Gigabit Ethernet IEEE 802.3Z Soon after the adoption of the specification ,IEEEE Set up a high-speed research group （High Speed Study Group,HssG), The research group is mainly committed to 10GE The study of .

10GE Not simply increasing the speed of Gigabit Ethernet 10 times ,2002 year 6 month ,IEEE802.3ae The committee formulated 10GE The official standard of , The standard mainly includes the following contents :

• compatible 8023 Minimum and maximum Ethernet frame lengths defined in the standard .
• Only full duplex mode is supported .
• Use point-to-point links and structured cabling to form a Star LAN .
• stay MAC/PLS Implement on the service interface 10Gbs The speed of .
• Define two physical layer specifications , LAN PHY And wan PHY.
• The definition will MAC/PLS The data transmission rate corresponds to the WAN PHY Adaptation mechanism of data transmission rate .
• Define interfaces that support specific physical media (PMD) Physical layer specification , It includes multimode fiber, single-mode fiber and corresponding transmission distance ; Support ISO/C11801 Types of optical fiber media defined in the second edition, etc .
• adopt WAN Interface sublayer ,10Gbi/s It can also be adjusted to a lower transmission rate .

Besides ,10Gbis No longer use copper wire , Only optical fiber is used as the transmission medium ; No longer use CSMA/CD agreement . Gigabit Ethernet can still use existing Fibre Channel Technology , but 10GE Use the newly developed physical layer .10GE Common physical layer specifications are as follows :

1. 10G Base-SR：SR Express " Short Reach"( short distance ),10GBae-SR For short distance connections only , The specification supports the encoding method of 64B/66B Short wave (850mm) Multimode fiber , The effective transmission distance is 2m~300m.
2. 10G Base-LR：LR Express “Long Reach”( The long distance ),10 G Base-LR It is mainly used for long-distance connection , The specification supports the encoding method of 64B/66B The long wave of (1310mm) Single mode fiber , The effective transmission distance is 2m~10km, Up to 25km.
3. 10G Base-ER：ER Express “ Extended Reach”( Super long distance ,10G Base-ER Support ultra long wave （1550nm) Single mode fiber , The effective transmission distance is 2m~40km.

3.4 Extended Ethernet

In most cases , We hope to expand the coverage of Ethernet . This section discusses the expansion of Ethernet at the physical layer and data link layer , This extended Ethernet is still a network in the network layer .

3.4.1 Extending Ethernet at the physical layer

The distance between hosts on Ethernet cannot be too far , Otherwise, the signal sent by the host will be attenuated to CSMA/CD The protocol doesn't work . So we can consider expanding Ethernet from logistics , To break through the limitation of distance .

3.4.1.1 Using optical fiber expansion

The host uses fiber and a pair of fiber modems to connect to the hub , It's easy to connect the mainframe to a hub a few kilometers away

3.4.1.2 Using hub extensions

Connect multiple Ethernet segments into larger ones through one or more hubs 、 Ethernet with multi-level star structure , For example, a department has a department , The second system , The third series is extended through the hub .

When introducing the advantages and disadvantages of this method , First, let's introduce what collision domain is .

• Collision domain （collision domain） It is also called conflict domain , It refers to the part of the network where the frame sent by one site will collide with the frame sent by other stations . There are three collision domains on the left of the above figure , On the right is a larger collision domain .
• The larger the collision domain is , The higher the probability of collision （ This involves CSMA/CD）.

It is obvious from the picture that , Advantages and disadvantages of hub expansion .

• advantage ：
  1. So that the computers on Ethernet belonging to different collision domains can communicate across collision domains
  2. Expand the geographical coverage of Ethernet
  3. The bridge makes each network segment an isolated collision domain
     
• shortcoming ：
  1. The collision domain has increased , But the overall throughput has not improved
  2. If different collision domains use different data rates , Then you can't use hubs to interconnect them

3.4.2 Extending Ethernet in data link layer

A more common way to expand Ethernet is at the data link layer . Early use of bridges , Now use Ethernet switches .

3.4.2.1 Use the bridge

The bridge works at the data link layer , It is based on MAC The destination address of the frame forwards the received frame . The bridge has the function of filtering frames . When the bridge receives a frame , This frame is not forwarded to all interfaces , But first check the purpose of this frame MAC Address , Then determine which interface to forward the frame to

The advantages and disadvantages of using a bridge are as follows ：

• advantage
  1. Filter traffic , Expand the scope of Physics , Improved reliability .
  2. Can interconnect different physical layers 、 Different MAC Sublayers and different rates （ Such as 10Mb/s and 100Mb/s Ethernet ） LAN .
• shortcoming
  1. Store and forward increase delay , stay MAC The sublayer has no flow control function , It's different MAC When the network segments of the sublayer are bridged together, the delay is larger .
  2. It is not possible for many to many hosts to communicate at the same time , Because the bridge can only analyze and forward one frame at a time .
  3. Bridge is only suitable for a small number of users ( No more than a few hundred ) And the LAN with less traffic , Otherwise, network congestion may occur due to the excessive broadcast information . This is the so-called broadcast storm .

Bridges are different from hubs ：

1. When the hub forwards frames , No detection of transmission media .
2. The bridge must execute before forwarding frames CSMA/CD Algorithm , If there is a collision in the transmission process , You have to stop sending and back off .
3. The hub works on the physical layer , The bridge works at the data link layer .

3.4.2.2 Use the switch

1990 The switched hub came out in , Many bridges are eliminated . Switching hubs are often called Ethernet switches or layer-2 switches .

Ethernet switch features ：

• Isolate collision domain
  Ethernet switch is essentially a multi interface bridge , Each interface is directly connected to the host , And it works in full duplex mode , The switch can connect many pairs of interfaces at the same time , So that each pair of hosts communicating with each other can act as an exclusive communication medium , Transmit data without collision , Ethernet switch uses a special switch structure chip , Forwarding by hardware is much faster than that by software .
  

• Exclusive bandwidth
  Ethernet with hub expansion , Divide the bandwidth equally , Ethernet extended by switch , Each site enjoys exclusive bandwidth , Virtually increased the total bandwidth .
  

• Don't use CSMA/CD
  Logically , Ethernet with hub expansion , The essence is bus Ethernet , You have to use CSMA/CD Protocol to handle collisions , And it can only work in half duplex mode , The switch adopts star Ethernet , Do not use shared bus , There is no collision problem , No need to use CSMA/CD, And it works in full duplex mode , However, the Ethernet frame structure is still used
  

The way Ethernet switches exchange frames ：

• Store and forward mode ： The whole data frame is first cached and then processed
• Straight through (cut-through) The way ： When receiving the data frame, immediately press the purpose of the data frame MAC The address determines the forwarding interface of the frame , Therefore, the frame forwarding speed is improved , The disadvantage is that it forwards the frame directly without checking for errors , Therefore, it is possible to forward some invalid frames to other sites .

Self learning function of Ethernet switch ：
The forwarding of switches still depends on Forwarding , Ethernet switches run self-learning algorithms to automatically maintain and publish .