What is Packet Switching?

In the communication process, the two communication parties use packets as the unit and use the store-and-forward mechanism to implement data exchange, which is called packet switching (PS: packet switching).

Packet switching is also known as packet switching. It divides user communication data into multiple smaller equal-length data segments, and adds the necessary control information in front of each data segment as the header of the data segment, each with a header. The data segments constitute a group. The header specifies the address to which the packet is sent. After the switch receives the packet, it will forward the packet to the destination based on the address information in the header. This process is packet switching. A communication network capable of packet switching is called a packet switching network.

The essence of packet switching is store and forward. It temporarily stores the received packets and queues them in the destination direction. When it can send information, it sends the information to the corresponding route to complete the forwarding. The store-and-forward process is the process of packet switching.

The idea of packet switching comes from message exchange. Message exchange is also called store-and-forward exchange. The essence of their exchange process is store-and-forward. The difference is that the smallest unit of information of packet exchange is packet, and message exchange is a Messages. Because transmission and exchange are performed in smaller packets, packet exchange is faster than message exchange. Message exchange is mainly used in public telegraph networks.

Chinese name: Packet switching
Foreign name: Packet Switching
Meaning: Transport and exchange in groups

Attributes: Store-and-forward exchange method
basic component: Consists of a packet header and subsequent user data
Exchange disadvantage: Delay and high overhead

Packet switching basic information

The essence of packet switching is to divide the data to be transmitted into many groups according to a certain length. In order to accurately transmit to the other party, each group is marked. Many different data packets are transmitted on the physical line in a dynamic sharing and multiplexing manner. Able to make full use of resources, when data packets are transmitted to the switch, they will be temporarily stored in the switch's memory, and then according to the busyness of the current line, the switch will dynamically allocate appropriate physical lines and continue the transmission of data packets until it is transmitted to the destination . After reaching the destination, the data packets are recombined to form a complete data.

A packet is composed of a packet header and a subsequent user data portion. The packet header contains the receiving address and control information. Its length is 3--10B, and the length of the user data part is fixed, with an average of 128B and a maximum of 256B. There is a problem that needs to be explained here: the packet length in the same packet network is fixed, but the packet length of different packet networks can be different. Packet switching: After the output port and the next node are determined by routing, the packet must be transferred from the input port to the output port using switching technology to achieve the transmission of bits through the network node.

Packet switching technology is developed to a certain degree in computer technology. In addition to making direct phone calls, people use computers and terminals to implement computer-to-computer communication. When the quality of transmission lines is not high and the network technology is relatively simple, it should be applied. And a kind of exchange technology.

Packet switching is also called packet switching. It divides the data transmitted by the user into multiple smaller equal-length parts, and each part is called a data segment. A packet is formed by adding some necessary control information to the front of each data segment. The header is used to indicate the address to which the packet is sent, and then the switch forwards them to the destination according to the address tag of each packet. This process is called packet switching. The communication network that performs packet switching is called a packet switched network. Packet switching is essentially developed on a "store-and-forward" basis. It has the advantages of both circuit switching and message switching.

In the packet switching mode, since the temporary storage and exchange of data can be performed in a packet mode, after processing by the switch, it is easy to implement communication between terminals of different rates and different procedures.

Development history of packet switching

Around 1970, people began to study a new long-distance digital data communication architecture: packet switching.

Packet switching Although the packet switching technology used has made great progress compared to that time, the basic technology of packet switching today is basically the same as the network technology of the 1970s, and packet switching is still a few effective ways to achieve long-distance data communication One of the technologies. And the two latest WAN technologies: Frame Relay and ATM are basically variants of packet switching.

Packet switching was developed as a technology for solving interactive processing applications. It is designed to support the transmission of bursty data streams. This type of service stream has a long continuous connection time and low service volume. The packet switching network uses statistical multiplexing technology, that is, multiple session connections can share a communication channel, which undoubtedly greatly improves transmission efficiency. However, shared communication links introduce latency. Therefore, a key issue we need to consider in the future is how the packet-switched network transmits delay-sensitive service flows, such as real-time service flows. In a packet-switched network, packets are routed through a series of intermediate nodes, usually across multiple networks. They are forwarded between a series of packet switches (ie routers) in a store-and-forward manner, and finally reach their destination. During transmission, the information is divided into packets containing the destination address and sequence number.

Packet switching classification

According to the implementation, packet switching can be divided into datagram packet switching and virtual circuit packet switching.

Packet switched datagram

Data packet switching requires at least one data transmission path between two communicating parties. The sender needs to prepare the data packets to be transmitted before communication. The data packets contain the address information of the sender and the receiver. Data packet transmission is independent of each other and does not affect each other. It can reach the destination according to different routing mechanisms and reassemble.

In this way, each packet is appended with a source and

Packet switching Information such as the destination address, packet number, packet start, end flag, and error check are transmitted in the form of packets in the network. The network only tries its best to deliver the packets to the destination host, but does not guarantee that the transmitted packets will not be lost, nor does it guarantee that the packets can reach the receiving end in the order in which they were sent. Therefore, the services provided by the network are unreliable and do not guarantee the quality of service. As shown in Figure 9-2 (a), some packets sent by host H1 to H5 pass through node ABE, some pass ACE or ABCE, and packets sent by host H2 to H6, some pass node BDE, and some pass BE . The datagram method is generally applicable to shorter single packet messages. The advantage is that the transmission delay is small, and when a node fails, it will not affect the transmission of subsequent packets. The disadvantage is that each packet has a lot of additional control information, which increases the length and processing time of the transmitted information, and increases the overhead.

Packet switched virtual circuit

The difference between it and the datagram method is that before the information is exchanged, a logical connection needs to be established between the sender and the receiver, and then packets are transmitted. All packets are exchanged and forwarded along the same path, and then dismantled after the communication is completed. The logical connection. The network guarantees that the transmitted packets reach the receiver in the order they were sent. Therefore, the services provided by the network are reliable and the service quality is also guaranteed. As shown in Figure 9-2 (b), all packets sent by host H1 to H5 pass through the same node ABE, and all packets sent by host H2 to H6 also pass through the same node BE.

This method is not suitable for users with high information transmission frequency and small transmission volume, but since each packet header only needs to be marked with the virtual circuit identifier and sequence number, the packet header has low overhead and is suitable for long message transmission.

Virtual circuit packet switching is like circuit switching. The two communicating parties need to establish a connection, but it is different from circuit switching. The packet switching connection is a virtual connection (also called a virtual circuit). There is no exclusive physical line in the connection. According to the implementation of the virtual connection, virtual circuits can be divided into switched virtual circuits and permanent virtual circuits.

Switching virtual circuits requires the two communicating parties to establish a temporary connection through a request and then communicate. When the communication ends, the temporary connection is removed.
The permanent virtual circuit is that the two communication parties do not need to request, but only need to establish a connection according to the agreement between the two parties and keep it for the agreed time.

From this, the characteristics of connection-oriented and connectionless operation are obtained.

(1) Characteristics of connection-oriented work

Regardless of whether it is a physical-oriented connection or a logical-oriented connection, its communication process can be divided into three phases: connection establishment, information transmission, and connection removal.

Once the connection is established, all the information of the communication is transmitted along this link path, and the orderliness of the information is guaranteed (the order of sending information is the same as the order of receiving information).

The delay of information transmission is smaller than that of connectionless operation.

Once the established connection fails, the transmission of information will be interrupted, and the connection must be re-established, so it is sensitive to failure.

(2) Characteristics of connectionless working mode

There is no connection establishment process, while routing and transmitting information again.

The information belonging to the same communication reaches the destination along different paths. The path cannot be predicted in advance, and the orderliness of the information cannot be guaranteed (the order of sending information and the order of receiving information are not consistent).

The delay of information transmission is larger than that of connection-oriented operation.

Not sensitive to network failures.

Packet Switched Packet Switched Features

(1) The smallest unit of information transmission is packet

The packet consists of a group header and user information. The packet header contains routing and control information.

(2) Connection-oriented (logical connection) and connectionless working modes

The virtual circuit adopts connection-oriented working mode, and the datagram is connectionless working mode

(3) Statistical time division multiplexing (dynamic allocation of bandwidth)

The basic principle of statistical time division multiplexing is to divide time into time slices of different lengths. The time slices of different lengths are the time required to transmit packets of different lengths. There is no fixed time slice allocation for each communication, but it is used on demand. This means that the length of the packet transmission time is transmitted using this multiplexed line, which shows that statistical time division multiplexing dynamically allocates bandwidth.

(4) Information transmission is error-controlled

Packet switching is a switching method specifically designed for data communication networks. Data services are characterized by high reliability requirements and no high-speed telephone communications for real-time performance. Therefore, in order to ensure the reliability of data information, CRC check is provided in packet switching. , Retransmission and other error control mechanisms to meet the needs of data business characteristics.

(5) Information transmission is not transparent

Packet switching processes the transmitted data information, such as splitting and reorganizing information.

(6) Flow control based on call delay system

In packet switching, when the data traffic is large, packets are queued for processing instead of being immediately lost as in circuit switching, so its flow control is based on call delay.

Packet switching advantages

Packet switched networks have many advantages over circuit switched networks

Packet switching

High packet switching utilization

Compared with the exclusiveness of circuit switching on the link, different data packets can be transmitted on the same link in a dynamic sharing and multiplexing manner, and the communication resource utilization is high, so that the channel capacity and throughput are increased. Great improvement. Because a single node-to-node link can be dynamically shared by many packets. Packets are queued and transmitted on the link as fast as possible.

Packet switched data rate

A packet-switched network can perform data rate conversion: two stations with different data rates can exchange packets, because each station is connected to this node at its own data rate.

Packet switching queuing mechanism

Data of different types and specifications can be transmitted on the same link at the same time. When there are a large number of packets on the packet network, a queuing mechanism for data transmission can be set to ensure that packets with higher priority are transmitted preferentially. When the circuit switched network is heavily loaded, some calls are blocked. On packet-switched networks, packets are still accepted, but their delivery delays increase.

Packet switching priority

When using priority, if a node has a large number of packets queued for transmission, it can transmit high-priority packets first. These packets will therefore experience less delay than lower priority packets.

Disadvantages of packet switching

Compared with circuit-switched networks, packet-switched networks also have some disadvantages.

Packet switching delay

Delay occurs when a packet passes through a packet-switched network node, while in a circuit-switched network

Packet switching There is no such delay.

Packet switching delay jitter

Because the packets between a given source station and destination station may have different lengths, they may take different paths, or they may experience different delays in the switches along the way, so the total delay of a packet may vary widely. Big. This phenomenon is called jitter. Jitter is undesirable for some applications (for example, in real-time applications such as phone voice and real-time images).

Large packet switching overhead

To transmit a packet across the network, the overhead information including the destination address and packet ordering information must be added to each packet. This information reduces the communication capacity that can be used to transport user data. In circuit switching, these overheads are no longer needed once the circuit is established. In addition, the packet-switched network is a set of distributed packet-switched nodes. In an ideal case, all packet-switched nodes should always know the status of the entire network. Unfortunately, because the nodes are distributed, there is always a delay between a change in the state of one part of the network and the other part of the network learning about the change. In addition, there is a certain cost involved in transmitting status information, so a packet-switched network will never operate "completely ideally".

Packet Switching Principle

There are two types of terminals hanging on the packet switching network: packet terminals and general terminals. The so-called packet-type terminal sends and receives information in the form of packets; in general, the information it sends and receives is a message, and it can only access the packet-switched network after being assembled and disassembled. If the sending terminal is a general terminal, the packet sent by PAD is divided into several packets and then sent to the packet switching network for transmission; if the receiving terminal is a general terminal, the PAD will re-set several packets belonging to a message. Assemble the message and send it to the general terminal.

The basic principle of packet switching is to use "store-and-forward" technology. When sending a message from the source station, the message is divided into packets with a fixed format (Packet), the destination address is added to the packet, and then the switch in the network After receiving the packet from the source station, it is temporarily stored in the memory, and then according to the provided destination address, it is continuously forwarded through other switches in the network to select an idle path, and finally sent to the destination address. This solves the communication between different types of users, and does not need to establish a physical path in the transmission process for a long time like circuit switching, but can be multiplexed in units of packets on the same line, so it is greatly improved The utilization of the line.

Packet switching is a switching technology that combines the advantages of circuit switching and message switching. The circuit switching process is similar to making a phone call. When the user needs to send data, the calling party needs to call and the switching network completes the called party before establishing a physical connection data path with it. When the connection needs to be disconnected, it is done by either of the communicating parties. . Its characteristics are suitable for sending large batches of information at one time. Due to the long connection establishment time, it is inefficient to transmit short packets. And the two parties are completely compatible in terms of information transmission rate, coding format, and communication protocol, which limits users of both sides with different rates, different coding formats, and different communication protocols to communicate. The basic principle of message exchange is to use "store-and-forward" technology. When sending a message from a source station, the destination address is added to the message, and then the switch in the network temporarily stores the message from the source station and stores it in memory. Then, according to the provided destination address, it will continue to choose the idle path for forwarding through other switches in the network, and finally send it to the destination address. This solves the communication between different types of users, and does not need to establish a physical path in the transmission process for a long time like circuit switching, but can be multiplexed on the same line in units of messages, so greatly Improved line utilization. However, this method has a long delay and large delay variation, and is not suitable for real-time and conversational communication, but it is suitable for e-mail, computer files, public telegram and other services.

Packet switching still uses the "store-and-forward" technology, but instead of exchanging messages in units of messages like message exchanges, it divides messages into packets with a fixed format (Packets) for exchange and transmission, generally 1kbit ~ Thousands of bits, each packet is appended with a source and destination address, packet number, packet start, end flag, error check and other information in a certain format and transmitted in the form of packets. After the source DTE transmits the packet to the local packet switch PSE in the form of a bit string, the local PSE receives the forwarding information required by each packet, regardless of whether the destination address device is connected, first stores it, then checks the destination address, and saves it in the PSE. The routing path specified by the destination address is found in the routing table of the. The PSE forwards the packet at the maximum allowed transmission rate. Similarly, each transit PSE stores and forwards each packet in this way until the packet is sent to the destination pSE, and then the PSE reaches the destination address DTE (see Figure 8-7). What is transmitted in the above manner is the datagram mode in packet switching. Generally suitable for shorter single packet messages. Its advantages are high transmission reliability and small transmission delay. As the memory capacity on the PSE is reduced, the economy is improved. The disadvantage is that each packet has more control information, which increases the length and processing time of the transmission information. Big overhead.

Another method of packet switching is called the virtual circuit method. The difference from the datagram method is that the source DTE sends a packet of a specific call request to the local PSE before the information exchange, which contains the address of the destination DTE and the logical channel identification. And forwarded by the PSE. If the call is accepted by the destination DTE, the corresponding response "call acceptance" will be answered, and the network will send a "call connection" to the source DTE. At this time, the call is established and a logical path called a virtual circuit is established between the two DTEs. Information can be transmitted on this virtual circuit until the end of data exchange, the virtual circuit is removed, and the corresponding logical channel identifier is released. Therefore, the virtual circuit method has three phases of virtual circuit establishment, data transmission, and dismantling at each communication, similar to the circuit switching method, but the transmission in the network is a packet switching method. This method is not suitable for users with high information transmission frequency and small transmission volume, but since each packet header only needs to be marked with the virtual circuit identifier and sequence number, the packet header has low overhead and is suitable for long message transmission.

Packet- switched protocol

Public packet-switched data networks

Packet switching An important public communication platform for long-distance data transmission between computers is a wide-area connection method commonly used internationally. The X.25 protocol developed by the International Telecommunication Union's telecommunications standards department ITU-TSS is an international standard supported and followed by many telecommunications organizations and manufacturers in the world. X.25 network is a public data network widely used internationally. X.25 was introduced by Tyltmet in 1970 and is the first generation of packet switching systems. The X.25 network was developed to transmit data, so it is not a direct competitive relationship with telephone service providers. X.25 packet switching technology emerged to meet the needs of interactive services. Interactive processing appeared in the late 1960s. It is a bursty data stream with long connection time but low data volume. X.25 provides a technology that enables multiple sessions to share the same communication channel.

X.25 Network When X.25 technology was proposed, the network was mainly an analog environment.

Packet switching A more serious problem with analog networks is that noise is amplified as it passes through some amplifiers, which can lead to very high error rates. Therefore, a value-added service provided by X.25 is to implement the error control function in the network. Because packet switching is a store-and-forward technology, at each intermediate node, error detection is performed on the packet. If all the packets are normal, the intermediate node will send a confirmation message to the original sending node; if there is an error in the packet received by the intermediate node, the node will send a message requesting retransmission. Therefore, at any node along the packet path, if noise is found to cause errors, errors will be corrected to ensure that the data stream is sent more accurately. Just like the PSTN network, the X.25 packet-switched network is also hierarchical.

In X.25 packet-switched networks, packet switches are divided into two categories: (1) A type of switch is close to the user end. In fact, they are located at the access point of the network. The functions of such switches include routing, forwarding packets, and error detection and correction. In addition, in order to support the access of different types of computers, to complete the conversion between different protocols, or between different working speeds and different encoding methods, some additional functions are needed, and these functions are at the access point. The switch is done. In other words, because it can perform the necessary conversion work according to your requirements and treat this conversion as part of the network business, so through the X.25 network, you can achieve the connection between different types of devices. (2) The second type of switches are inside the network. They do not provide the aforementioned high-level value-added functions. They only complete packet routing, forwarding, and error control and detection. Non-standard X.25 terminals need to be connected to an X.25 network through a packet disassembly device (PAD). PAD completes the conversion of the protocol and generates the packets specified by the X.25 standard, so that the data can be transmitted through the X.25 network. These PADs can be placed on the user side or on the network side.

Packet switching X.25 is essentially a standard access protocol between user equipment and packet-switched networks established by ITU-T. It defines the interface for terminals in packet mode to access the public data network through dedicated circuits. There are also some commonly used X-series protocols, these are:

(1) X.28 is a standard protocol between terminal equipment and PAD;

(2) X.29 is a standard protocol between PAD and the network;

(3) X.75 is a gateway protocol for interconnecting two or more packet-switched networks. One of the networks can be a dedicated packet data network and the other can be a public packet data network, or they are two different operators Network and more.

Advantages and disadvantages of X.25 network: judging the advantages and disadvantages of a particular technology depends on the environment in which it is located. In an analog information base environment, noise is a big problem, so error control is necessary. However, in each intermediate node, in addition to selecting a route to determine the next hop, error control must be implemented for each packet, which will increase the end-to-end transmission delay. Because the X.25 packet-switched network is only used to transmit data, delay or packet loss is not a critical parameter for it. Another contribution of X.25 is its packet size. It uses relatively small packets, typically 128 bytes or 256 bytes. The evaluation of the pros and cons of this issue also changes over time. Due to the noise factor, small packets are very advantageous in X.25 networks. If there is noise in the network, errors will occur, so the packet often needs to be retransmitted. Obviously, retransmitting relatively small packets is more efficient than retransmitting large chunks of information, so X.25 is specifically designed to use smaller packets. Also, because it is an early network, it is suitable for links operating at relatively low rates. The link rate ranges from 56kbit / s to 2Mbit / s.

The advantages of X.25 are as follows: (1) Because X.25 is the first to provide Layer 3 network address information

Packet switching , So that the packet can be routed and relayed in a series of intermediate nodes and networks, so it has a strong addressing function; (2) because of the use of statistical multiplexing technology, its bandwidth utilization is high; (3) The packet can bypass the congested node and re-route through other connections and nodes, thus improving the congestion control capability; (4) can continuously detect and correct all types of errors on each intermediate node Therefore, the error control function is improved; (5) When nodes and lines fail, the routing can be re-selected, so the availability is very high.

The disadvantages of X.25 are as follows: (1) large queuing delay; (2) low-speed communication links; (3) smaller packet sizes and lower bandwidth utilization than the new protocols using larger size packets; There is no QoS guarantee, so it is not suitable for delay-sensitive applications; (5) It is only used to transmit data, and today we are trying to find a comprehensive service solution.

X.25 protocol introduction: An important part of using public data networks is the interface to them. ITUX.25 standard is a widely used interface. Many people use the term "X.25 network", which has led many to mistakenly believe that X.25 defines a network protocol. But this is not the case. X.25 only defines the protocol between DTE and DCE connected to the public data network (Figure 9-4). Therefore, X.25 can be strictly used as a user-network interface or a user-user interface through a public data network. The X.25 protocol refers to the interface between the data terminal equipment (DTE) and the data circuit terminal equipment (DCE) used by the terminals that work in a packet mode and are connected through a dedicated circuit and a public data network. It defines three layers of protocols: physical layer, data link layer, and packet layer (ie network), which correspond to the lower three layers of the ISO / OSI seven-layer model.

(1) Physical layer: The basic function is to establish, maintain, and tear down the physical link between DTE and DCE, defining the physical

Packet switching The mechanical, electrical, functional and regulatory characteristics of the management link provide synchronous, full-duplex point-to-point bitstream transmission means. The interface between the DTE and the local DCE is specified in accordance with X.21. (2) Data link layer: Provides the packet layer with a retransmission waiting and error control method through the physical link between the DTE and the local packet switch PSE (PacketSwitchedEquipment), so it has high reliability. The LAPB specified in this layer The (LinkAccessProcedureBalanced) procedure is a balanced subset of the HDLC procedure. It mainly specifies the establishment and dismantling procedures of the data link, the established information transmission procedures, and error control and flow control. In addition, this layer also specifies the Multilink Procedure (MLP), which transmits information frames on multiple parallel data links simultaneously to improve the throughput and reliability of information.

(3) Packet layer (network layer): mainly describes the packet layer procedures for exchanging control information and user data on the DTE / DCE interface, and specifies the virtual circuit business procedures, basic packet structure, data packet format, and optional user service functions. . This layer uses the principle of time division multiplexing to realize that a source DTE uses a physical circuit to call multiple destination DTEs for packet data exchange. In addition, it also provides a permanent virtual circuit PVC service. This is a virtual circuit for fixed use by users. The source DTE can use the virtual circuit without having to establish a call. The interrelationship of the layered protocols in X.25.

Packet Switched Network Composition

The packet-switched network structure is generally composed of

Packet switching It consists of basic equipment such as group switch, network management center, remote concentrator, assembly and disassembly equipment, packet terminal / non-packet terminal and transmission line. (1) The packet switch implements the interface protocol (X · 25) between the data terminal and the switch, the signaling protocol (such as X · 75 or internal protocol) between the switches, and stores and forwards in packets, providing packet network services Support, cooperate with the network management center to complete routing selection, monitoring, billing, control, etc. According to the status of the packet switch in the network, it is divided into two types: transfer switch and local switch; (2) The network management center (NMC) and the packet switch cooperate to ensure the normal operation of the network.

Its main functions include network management, user management, measurement management, billing management, operation and maintenance management, routing management, collecting network statistics and necessary control functions, etc., which are the core of the entire network management; (3) points The main function of the assembly and disassembly device (PAD) is to convert the non-packet format of ordinary character terminals into a group format, and group the data streams of each terminal into groups. The groups are interleaved and multiplexed on the collective channel. The format is converted in the opposite direction. (4) The function of the remote concentrator is similar to that of a packet switch, and usually contains the function of a PAD. It is only connected to a packet switch and has no routing function. It is used in areas where users are concentrated, and is generally installed in the telecommunications sector. (5) Provide basic network services: switching virtual circuits and permanent virtual circuits, and other supplementary services, such as closed and user groups, and network user identification. When communicating between end-to-end computers, routing and flow control are performed. Can provide a variety of communication procedures, data forwarding, maintenance operation, fault diagnosis, billing and some network statistics.