OSI Seven layer reference model

machine language ： use 0,1 To represent electrical signals

Abstract language ： written words , voice , picture , Video and other things that human beings can understand

The conversion process ： Abstract language ——> code ——> Binary system ——> Electrical signals

By adopting protocols （ A standard established ） Use a certain application to obtain a certain service

application layer Provide all kinds of services , Convert abstract language to coding , Provide human-computer interaction interface

The presentation layer Convert the encoding to binary , Unified format

The session layer Maintain session links between web applications and web servers —— There is no standard （ for example ：qq Number , Game nicknames and so on ）

The top three , Process parts of the data for the application , Collectively referred to as the data application layer , Write the program on the upper three levels

Transport layer Realize end-to-end communication （ Application to application communication , Differentiate applications ）

Port number 16 Bit binary structure ,0~65535, True for 1~65535. among 1~1023 Indicates the port （ Static port ） Used to calibrate known application services ;1024-65535 Dynamic port , High port

When the terminal communicates with the server , Use the random high port to correspond to the local process number , Use the indicated static port to mark the specific service accessing the server . The core function of the port number is to distinguish the process of the local device and the specific services of the server

Some common port numbers

FTP File transfer protocol TCP 20（ To transmit data , file ）/21（ Transmission command, etc ）

SSH Telnet Protocol TCP 22

telnet Telnet Protocol TCP 23

DNS Domain name resolution protocol TCP/UDP 53

DHCP Dynamic Host Configuration Protocol UDP 67（ Server port ）/68（ Client port ）

TFTP Simple file transfer protocol UDP 69

HTTP Hypertext transfer protocol TCP 80

HTTPS HTTP+SSL（ Secure encryption protocol ）TLS TCP 443

Data segmentation

Data segmentation （ suffer MTU Limit ,MTU The data transmission unit is , Maximum 1500）, Network transmission is intermittent , When data packets become continuous, it will affect other discontinuous packets , The delay gets higher , The packet is broken and needs to be retransmitted , The big bag needs to be segmented

The network layer Use IP Address to realize logical addressing between hosts

Get the target IP Methods

1, Know the server directly IP Address

2, Access the server through the domain name

3, Access the server through the application

4, Get... Via broadcast

IP Address

IPV4——32 Bit binary structure （ dotted decimal ） IPV6——128 Bit binary structure （ It's hexadecimal ）

Distinguish between different broadcast domains

Network bit The network bits are the same , It means that it is in the same broadcast domain

Host bit In the same broadcast domain , Use the host bit to distinguish different hosts

Subnet mask to distinguish between network bits and host bits , Continuous 1 And continuous O form ,1 On behalf of network bit ,0 Represents the host bit

A,B,C For unicast address （ Can be source IP Or target IP）

A：255.0.0.0 B：255.255.0.0 C：255.255.255.0

D For multicast address （ As a goal IP）

E To keep the address

A：0XXX XXXX(0~127) 1~126 B：10XX XXXX(128-191) C：110X XXXX(192~223) D：1110 XXXX(224~239) E：1111 XXXX(140~255)

Unicast is one-to-one , Multicast is one to many （ The same multicast group ）, Broadcast a bunch of all （ All in the broadcast domain ）

stay IP In the address space , Part of it IP The address is called private IP Address （ The private network IP Address ), The rest is called public IP Address （ Public network IP Address )

A: 10.0.0.0 - 10.255.255.255 amount to 1 individual A Class segment

B: 172.16.0.0 - 172.31.255.255 amount to 16 individual B Class segment

c: 192.168.0.0 - 192.168.255.255 amount to 256 individual c Class segment

special IP Address

127.0.0.1 -127.255.255.254 Loopback address （ Test whether there is a problem at the software level ）

255.255.255.255 Restricted broadcast address （ Restricted by router , Only as a target IP）

All the hosts 1 192.168.1.255/24 Direct address （ Only as a target IP）

All the hosts 0 192.168.1.0/24 Network segment

0.0.0.0 No, IP Address ; all IP Address

169.254.0.0/16 Local link address / Automatic private address

VLSM Variable length subnet mask

CIDR No inter domain routing

Take the same , Go different

HYPERNET The subnet mask is greater than the type subnet mask

Summary The subnet mask is smaller than the type subnet mask

Data link layer Control physical hardware , Convert binary to electrical signal , Need to pass through MAC Address for physical addressing ,48 Bit binary structure （ Global unity , use the same pattern ）

Get the target MAC Address method （arp agreement （ Address resolution protocol ）

• positive ARP adopt IP The address for MAC Address

  The host sends... In the form of broadcast ARP request , Based on the known IP Address , request MAC Address , All devices in the broadcast domain can receive request packets , All devices that receive packets will record the source in the packet IP Address and source MAC The correspondence of the address is related to the local ARP The cache table . after , Look at the request IP Address , If it's local IP Address , Will be performed ARP The reply , Local MAC Address inform . If it's not local IP Address , The packet will be discarded directly . after , When sending data again , Give priority to local ARP Cache table , If there are records in the table , Then the information will be sent directly according to the record , If there is no record , Then send ARP Request package acquisition MAC Address .

• reverse ARP adopt MAC The address for IP Address

• free ARP Use positive ARP How it works ask yourself IP Address （ Check for address conflicts and let others get their own IP Address ）

The physical layer Process and transmit electrical signals

TCP/IP Model

CP/IP Protocol cluster （ The aggregation of rules . There is an agreement first , Then draw lessons from the idea of layering ）

application layer

The network layer

The network layer

Network interface layer （ Data link layer , The physical layer ）

TCP/IP The standard model —— Four layer model TCP/IP peer model —— Five layer model

PDU Protocol data unit

OSI in ：L1PDU......L7PDU（ From top to bottom ）

TCP/IP in ： The data packet , Data segment , Data packets , Data frame , Bit stream

Encapsulation and de encapsulation

application layer —— There is encapsulation in the application layer , Depending on the application

Transport layer —— Port number

The network layer ——IP Address

Data link layer —— Ethernet is MAC Address

The physical layer

TCP/IP Support cross layer encapsulation ,OSI Cross layer encapsulation is not supported

1. Cross four layer packaging Between routers OSPF（89）

2. Span three , Four layer package Between switches STP

Forerunner Frame header delimiter 802.2 Header and Data（LLC Logical link control layer ） contain DSAP（ Indicate what protocol the upper layer of the device receiving the data frame uses to process the data ）,SSAP（ Indicates the protocol generated by the upper layer of the device that sends the data frame ）,control（1. Connectionless mode 2. Connection oriented pattern . It can realize the segmentation and reorganization of data packets ）

Generally, it is used to send information between devices connected directly at a short distance , It can improve forwarding efficiency

Generation conditions of direct route ： Both physical level and protocol level UP; The interface must be configured IP Address

Get unknown route information of network segment

Static routing ： Routing entries manually configured by the network administrator

Dynamic routing : All routers run the same routing protocol , after , Routers communicate with each other , Exchange and calculate the routing information of the unknown network segment .

Static routing

advantage ： The choice of route is more reasonable ; safer ; No additional resources are required

shortcoming ： Large amount of configuration ; Static routing cannot automatically converge based on changes in topology

Basic configuration

Method 1 :[r1]ip route-static 192.168.3.0 24 192.168.2.2---- You need to recursively find the interface

Method 2 :[r1]ip route-static 192.168.3.0 24 GigabitEthernet 0/0/1 --- Write the interface directly [r2-GigabitEthernet0/0/0]arp-proxy enable --- The activation agent ARP

After activation , The router will check the received ARP Request package , be based on ARP Requested IP Address view local routing table , If the local route can reach , Will pretend to be the other party's IP Address reply ARP The reply . To his own MAC Address reply , after , The data is sent to the router , The router forwards .

Method 3 :[r1]ip route-static 192.168.3.0 24 GigabitEthernet 0/0/1 192.168.2.2 Recursive lookup is not required

Static route expansion

Load balancing

When a router accesses the same target with multiple paths with similar overhead , You can split the traffic and take multiple paths at the same time , Play the effect of superimposing bandwidth .

Manual summary

When a router can access multiple consecutive subnets , If they all pass the same next jump , These network segments can be summarized and calculated , Directly write a static route to the summary network segment . You can reduce the number of routing table entries , Improve forwarding efficiency .

Routing black holes

In summary , If the network segment that does not actually exist in the network is included , It may cause the flow to have no return , Waste link resources . Reasonable division and aggregation can reduce the generation of routing black holes

Default route

A route that does not limit the destination . When looking up the table , If all routes do not match , The default route will be matched .

[r1]ip route-static 0.0.0.0 0 12.0.0.2

Empty interface routing

Empty interface routing ---- Prevent routing black holes and default routes from meeting and leaving the ring . 1,NULL 0 --- If the outgoing interface of a route entry points to an empty interface , The datagram matching this route information will be discarded 2, Matching principle of routing table --- The longest match principle （ The principle of exact match , Match the longest subnet mask )

[r1]ip route-static 192.168.0.0 22 NULL o

Floating static routing

By modifying the default priority of static routing , Achieve the effect of backup

[r2]ip route-static 192.168.0.0 22 21.0.0.1 preference 61