TCP/IP is usually divided into four layers, as shown in:
Generally, the application layer program belongs to the user process and cares about the details of the program. The other three layers are executed in the kernel and focus on the communication details. The Application Layer Program does not care whether its host is on Ethernet or other types of networks. It only transmits data downward. Different types of physical networks are connected through routers, this enables applications on different networks to communicate with each other. The network layer provides unreliable services. TCP provides a reliable transport layer on the network layer, including timeout retransmission and validation grouping.
1. Where the application layer network applications and their application layer protocols are stored. Application layer protocols are distributed across multiple end systems. Applications in one end system exchange information with applications in another end system. Groups information at the application layer into messages.
2. Transport Layer provides the service for transmitting application-layer packets between application endpoints. TCP: Provides connection-oriented services (ensure delivery, traffic control, and congestion control) to applications ). It is a reliable transmission from one end system to another. UDP: provides applications with connectionless services (unreliable, no traffic control, and no congestion control ). The transport layer is called a packet segment. The Transport Layer Protocol needs to store a 16bit Application Identifier (that is, the port number) when generating the packet header.
3. The network layer is responsible for moving the network layer group called datagram from one host to another. The Internet Transport Layer Protocol (TCP/UDP) in the source host submits the transport layer packet segment and Destination Address to the network layer. IP protocol: defines fields in the datagram and how the end system and router act on these fields. Adhesive used to connect the internet. Routing Protocol: determines the route. The data unit sent from the IP layer to the link layer is called an IP data packet. Because TCP, UDP, ICMP, and IGMP all transmit data to IP addresses, a flag must be added to the generated IP address header to indicate which layer of data belongs.
4. The link layer sends the group from one node to another (the entire frame is moved from one network element to the adjacent network element ). Ethernet, Wi-Fi, and point-to-point (PPP) protocols. The link layer is grouped into frames (bytes ). Network Interfaces must send and receive IP, ARP, and RARP data respectively. Therefore, an identifier must be added to the Ethernet frame header to identify the type of data.
5. The physical layer moves a single bit in the frame from one node to the next node. For example, Ethernet has many physical layer protocols: About twisted-pair copper wires, about coaxial cables, and about optical fibers. The method for moving a bit across these links is different.
Transport Layer: Provides logical communication between processes. The active port number and destination port number are saved in the TCP packet. Network Layer: Provides logical communication between hosts. The active IP address and the destination IP address are saved in the IP datagram format.
The network layer provides point-to-point (between hosts) services, while the transport layer provides end-to-end (between processes) services. The most basic task of TCP and UDP: Expand the IP interaction service between the end systems to the interaction service between processes.
Each network interface on the Internet corresponds to an IP address. The five types of IP addresses are as follows:
The IP address ranges are as follows:
The server must specify the port so that the client can determine the port to connect. The server is identified by well-known port numbers, for example:
- FTP: tcp21
- TELNET: tcp23
- TFTP: udp69
The port occupied by the client after connecting to the server program is the available port (temporary port number) in the system automatically allocated by the system by default. This port number can be obtained by the server by calling the accept function. The client usually does not care about the port number it uses. Generally, the temporary port number ranges from.
For details, see chapter 1st TCP/IP details.
