Computer network: it is the product of the development of computer technology and communication technology, and it is developed with the requirements of society for information enjoyment and information transmission. The so-called computer network is a system that uses communication equipment and lines to interconnect multiple computer systems with different geographical locations and independent functions, and realizes resource sharing and information transmission in the network with well-functioning network software (i.e. network communication protocol, information exchange mode, network operating system, etc.). ).
Composition of computer network: It usually consists of three parts, namely, resource subnet, communication subnet and communication protocol.
Resource subnet: the user-oriented part of the computer network, which is responsible for the application-oriented data processing of the whole network. Its main body is all the main computers connected to the computer network, as well as the external equipment, software and data that these computers have for * * *.
Communication subnet: the part of a computer network that is responsible for data communication. The communication transmission medium can be twisted pair, coaxial cable, radio communication, microwave, optical fiber, etc.
Communication protocol: In order to make the communication between computers in the network reliable and effective, the rules and protocols that both parties must abide by are called communication protocols.
* * * Resources: including hardware and software resources. Hardware resources, such as high-performance processing components with special functions and high-performance input and output devices (laser printers, plotters, etc.). ) and large-capacity auxiliary storage devices (such as tape drives and large-capacity hard disks). ) can save hardware expenses. Software resources such as software and data.
Local area network (LAN): A communication system that allows several independent computers to communicate directly at an appropriate transmission rate within an appropriate range. General networks can be classified according to their size. Usually, most of the networks we use in the office or at home belong to the local area network. Because the distance between computers is short, there is no need to relay through too many network devices. This kind of network feels faster, but it also has a smaller scope of application.
WAN: Compared with LAN, anything beyond the scope of LAN can be regarded as WAN.
Metropolitan area network (MAN): A network that operates in a city or physically uses the basic telecommunication facilities of the city (such as underground cable system), which is sometimes different from the WAN, and is called MAN.
Network architecture: refers to the overall design of communication system, which provides standards for network hardware, software, protocol, access control and topology. It widely adopts the reference model of OSI- open systems interconnection proposed by the International Organization for Standardization (ISO) in 1979. The OSI reference model describes seven levels of network structure: physical layer, data link layer, network layer, transport layer, dialogue layer, presentation layer and application layer. Its specification is open to all manufacturers and has the function of guiding the international network structure and the trend of open systems. It directly affects the performance of bus, interface and network. Common network architectures include FDDI, Ethernet, token ring network and fast Ethernet. From the perspective of network interconnection, the key elements of network architecture are protocol and topology.
Protocol: It is a formal description of data formats and rules that must be observed when exchanging data between computers. Simply put, if computers in the network can communicate with each other smoothly, they must speak the same language, which is equivalent to protocols, including Ethernet, NetBEUI, IPX/SPX and TCP/IP protocols.
Topology structure: refers to the interconnection form between sites in the network, mainly including bus topology, star topology, ring topology and their mixed types.
FDDI/CDDI: formulated by X3T9.5 of ANSI. The rate is100 Mbps; CDDI is FDDI based on copper cable (twisted pair). FDDI technology is mature, and the network can be extended to 100 km. Because of its ring structure and excellent management ability, it has high reliability. Expensive, complex installation, perfect standards, mature technology and rich software and hardware products.
IEEE 802.5/ Token Ring Network: Commonly used in IBM systems, the supported rates are 4Mbps and 16Mbps. Novell and IBM LAN servers support 16MbpsIEEE802.5/ token ring technology.
Switched Ethernet: The supported protocol is still IEEE802.3/ Ethernet, but it provides multiple independent 10Mbps ports. It is fully compatible with the original IEEE802.3/ Ethernet, and overcomes the decrease of network efficiency caused by * * * enjoying 10Mbps.
100base-t Fast Ethernet: The difference with 10BASE-T is that the network speed is increased by 10 times, that is, 100M. The PMD protocol of FDDI is adopted, but the price is cheaper than FDDI. The standard of 10BASE-T is formulated by IEEE802.3. It adopts the same media access technology, similar step size rules and the same outgoing line as 10BASE-T, and is easy to integrate with10base-t. Only two repeaters are allowed in each network segment, and the maximum network span is 2 10.
IEEE 802.3/ Ethernet: the most extensive medium access technology, which usually runs in the physical layer and data link layer of the OSI model. It is the main media access technology adopted by Novell, Widows NT, IBM, UNIX LANServer, DECNET and other low-level organizations. Its networking mode is flexible and convenient, and it supports a variety of software and hardware products. Its rate * * * does not include 10Mbps. According to different media, it can be divided into: 10BASE-2 (coaxial thick cable), 10BASE-5 (coaxial thin cable), 10BASE-T (twisted pair) and 10BASE-FL (optical fiber).
NetBIOS/netbeui: NetBIOS is the industrial standard of LAN software interface, which can support a variety of transmission media. NETBEUI is an extended user interface of NETBIOS, which is adopted by Microaoft Windows NT and IBM LAN Manager. NETBIOS was developed earlier and simpler, without considering the interconnection between networks, and its naming scheme is not suitable for many operating systems.
IPX/SPX: the main protocol of Novell network. There are many hardware and software that support IPX/SPX and I/O devices. In the OSI reference model, it is equivalent to the third and fourth layers (network layer and transport layer). In NOVELL network, you can load the IP protocol NETBIOS protocol on IPX.
TCP/IP: IP is widely deployed in UNIX and has become a de facto international industrial standard. IP is also the main protocol of the Internet. IP protocol can span LAN and WAN, and almost all LAN and WAN devices support IP protocol, which is the best protocol to unify media transmission mode. IP protocol is a data protocol with better response time and less protocol interaction, which is more suitable for high-speed transmission.
Bus topology: a single transmission line is used as the transmission medium, and all stations are directly connected to the trunk cable, that is, the bus, through corresponding hardware interfaces.
Star topology: All sites are connected to a central point, which is called the hub of the network.
Ring topology: All stations are connected in series, just like a chain, forming a loop or ring.
Mixed topology: After the interconnection of residential networks, there will be some mixed forms of topological structures, that is, mixed topology.
Transmission medium: it is the physical path between the sender and the receiver in the communication network. Common network transmission media include twisted pair, coaxial cable and optical cable.
Twisted pair: it is the most commonly used transmission medium in integrated wiring system, especially in star network topology, twisted pair is an essential wiring material. Twisted pair includes one or more twisted pairs. In order to reduce the interference of signals, each twisted pair cable is generally formed by winding two insulated copper wires. Twisted pair can be divided into unshielded twisted pair (UTP) and shielded twisted pair (STP). Among them, STP can be divided into three categories and five categories, while UTP can be divided into three categories, four categories, five categories and more. At the same time, Category 6 and Category 7 twisted pairs will also be used in the wiring system of computer networks in the near future.
RJ-45 connector: Both ends of each twisted pair are connected to the network card and hub (or switch) by installing RJ-45 connector (commonly known as crystal head).
Coaxial cable: It consists of a hollow cylindrical netted copper conductor and a copper conductor located on the central shaft, and the copper conductor, the hollow cylindrical conductor and the outside are separated by insulating materials. Compared with twisted pair, coaxial cable has strong anti-interference ability and good shielding performance, so it is often used in the connection between devices or bus network topology. According to different diameters, it can be divided into thin cables and thick cables.
BNC connectors: BNC connectors are installed at both ends of thin cables and connected with network cards and hubs (or switches) through special T-connectors.
Optical fiber: optical fiber, that is, optical fiber, is a fine and flexible medium that can transmit optical signals. An optical cable is composed of multiple optical fibers. Compared with twisted-pair cable and coaxial cable, optical cable meets the requirements of long-distance transmission of large-capacity information in the network and plays a very important role in the computer network.
Half-duplex: It means that although the network card can send and receive data, it can only do one action at a time, and cannot send and receive at the same time.
Full duplex: This means that we can receive and send signals at the same time. For example, the telephone is a full-duplex transmission device, and we can also talk to each other while listening to each other. Theoretically, full-duplex transmission can improve network efficiency, but it is still useful in cooperation with other related devices. For example, twisted pair network cable must be selected for full duplex transmission, and the hub connected in the middle should also be able to transmit full duplex; Finally, the adopted network operating system must also support full-duplex operation in order to truly exert the power of full-duplex transmission.
Programmatic I/O: This is an effective transmission mode since its early use, which was adopted by NOVELL's NE 2000 network card which was popular all over the world at that time. The transmission efficiency of this transmission method is not easy to improve, and once it encounters a large amount of data, it becomes the bottleneck of transmission.
Shared memory: This network card puts the data to be transmitted into the memory on the card, and this memory must occupy an address in advance (most of the addresses occupied are between 640- 1024KB). With this address, this memory can be regarded as a part of the motherboard memory: when the host requests data from the network card, it will retrieve it directly from this memory; Conversely, putting data into memory is equivalent to transmitting data to the network card. If the programmed I/O mode is compared to using a spoon to scoop water, then sharing memory is to draw water with a bucket, which can highlight its efficiency in the case of large transmission volume.
Bus master control: This network card has a controller, which is specially used to control the whole transmission process and the use of the bus. Because the control action is completed by this chip, data can be directly transmitted from the network card to the motherboard without I/O PROT or CPU. Because it does not occupy valuable CPU time, it can effectively reduce the system burden, so it is especially suitable for servers. Most network cards with EISA, MCA and PCI interfaces support communication with the motherboard in this bus master mode.
802.3x flow control: More efficient data transmission improves performance. The network card establishes the best data transmission by communicating with the switch.
Parallel task technology: 3COM's patented technology enables the highest data transmission speed when 10Mbps or 100 Mbps is connected.
Parallel task 2 technology: 3COM's patented technology, because of more effective data transmission on PCI bus, can reduce CPU occupancy and improve application performance. In the past, in a bus main operation cycle, the network card only allowed 64 bytes of data to be transmitted on the PCI bus at most. In order to transmit a 15 14 byte data packet to the PC host, 24 separate bus main operation cycles are needed, which makes the bus very inefficient. Through the parallel task II technology, the network card can transmit the whole Ethernet packet to the bus in a main bus operation cycle, which greatly improves the efficiency of the PCI bus. In this way, the transmission speed is accelerated, the system performance is improved, and the application software of desktop and server works better.
32-bit bus master DMA: wide data path, high-speed transmission, low CPU occupancy, providing the best system performance.
Interactive access technology: the network card can dynamically analyze the network information flow, and then adjust the network performance.
Remote wake-up: enables the network administrator to command the remote PC to start at the central location, which is convenient for updating and maintaining the desktop after work hours (the PC motherboard must be equipped with a 3-pin remote wake-up connector; You also need desktop management application software, which can generate the remote wake-up signal of Magic Bag TM).
DMI2.0: Support remote computers to record and report computer status to improve desktop management.
3Com DynamicAccess software: it is an organic part of 3Com Fast EtherLink XL series, which adds various intelligences to network cards. Include 1, and differentiate the priority of data flow by service category. Assign high priority to time-critical data to improve the performance of multimedia and business-critical applications; 2. Distributed RMON(dRMON) intelligent agent TM software. The software can provide comprehensive and cheap network management in switched and high-speed network environment, including supporting various remote monitoring; 3. Fast IP software. This software alleviates the possible bottlenecks of routers to the greatest extent, thus improving the interconnection performance between networks. 4. Effective multicast control. This control can automatically filter out unnecessary multicast streams before they flood the LAN, thus expanding the useful bandwidth of the network.
100VG-AnyLAN: made by HP, AT & amp； T organization development, the standard is formulated by IEEE802. 12. Its advantages are that it can be based on three kinds of 8-core twisted pair networking, supports priority scheduling, is suitable for transmitting multimedia information, and is cheap. Disadvantages are immature standards, lack of fault-tolerant backbone, limited confidentiality and few supporting products.
ATM: A packet-based high-speed network, which will be the main communication transmission means of the future information superhighway. ATM standard is formulated by ATM Forum (more than 150 countries participate). Based on a 53-byte cell, the data exchange rate can reach 25M, 34M, 45M, 50M, 155M, 622M and several Gbps. ATM supports more and more products, but the price is higher.
Development history:
In 1980s, with the development of microcomputer technology, microcomputer LAN technology and products developed rapidly. At the end of 1980s, foreign microcomputer circles predicted that the environment used by microcomputers in 1990s was network. Actually, it's true. The development of personal computer network has a considerable impact on the whole computer network field. Thousands of computer network users are distributed in various application fields, which promotes the development of network application technology, thus accelerating the development of computer network technology.
In the past, foreign personal computer network products have always occupied the network market, among which NOVELL, 3COM, IBM, BANYAN and SUN products are the main users. With the development of network, manufacturers in Taiwan Province Province have developed rapidly with the advantages of strong production capacity and many factories in mainland China, and brands such as D-Link and TP-Link have gradually matured. In addition, domestic computer product manufacturers Shida and Lenovo also produced their own network products.
In fact, the development history of network card is also the development history of network. ....
Talking about network card;
Different classification of network cards: specially designed according to the working characteristics of different servers, which are more expensive but have better performance. As far as compatible network cards are concerned, network cards are generally divided into ordinary workstation network cards and server-specific network cards. Server-specific network cards are classified according to the following standards: 10M network card, 100M network card, 10/ 100M adapter network card and 1000M network card according to the bandwidth supported by network cards; According to the different bus types of network cards, they are mainly divided into three categories: ISA network cards, EISA network cards and PCI network cards, among which ISA network cards and PCI network cards are commonly used. The bandwidth of ISA bus network card is generally 10M, and that of PCI bus network card ranges from 10M to 1000M. The same is the 10M network card. Because the ISA bus is 16 bits and the PCI bus is 32 bits, the PCI network card is faster than the ISA network card.
Interface types of network cards: According to different transmission media, there are three interface types of network cards: AUI interface (thick cable interface), BNC interface (thin cable interface) and RJ-45 interface (twisted pair interface). Therefore, when choosing a network card, you should pay attention to the interface type supported by the network card, otherwise it may not be suitable for your network. The common 10M network cards in the market mainly include single-port network cards (RJ-45 interface or BNC interface) and dual-port network cards (RJ-45 and BNC interface), and there are few network cards with AUI thick cable interface. 100M and 1000M network cards are generally single-port cards (RJ-45 interface). In addition to the interface of the network card, we often pay attention to whether the network card supports diskless startup when choosing the network card. If necessary, consider whether the network card supports optical fiber connection.
Selection and purchase of network cards: According to statistics, most local area networks adopt Ethernet technology, so take Ethernet network cards as an example to discuss some problems that should be paid attention to when selecting network cards. Pay attention to the following points when buying:
Network card application: Ethernet cards include 10M, 100M, 10M/ 100M and Gigabit cards. For large-scale data networks, servers should use Gigabit Ethernet cards, which are mostly used for the connection between servers and switches to improve the response speed of the whole system. 10M, 100M and 10M/ 100M network cards are commonly used network devices, and the prices of these three products are not much different. The so-called 10M/ 100M adaptation means that the network card can automatically negotiate with the remote network equipment (hub or switch) to determine whether the current available rate is 10M or 100M. For common applications such as file sharing, 10M network card is enough, but for voice, video and other applications that may appear in the future, 100M network card will be more conducive to the transmission of real-time applications. For the existing foundation of 10M technology (such as hubs and switches, etc.). ), the usual workaround is to buy 10M/ 100M network card, which is beneficial to protect the existing investment and further expand the network. In terms of overall price and technical development, it will take some time for Gigabit Ethernet to reach the desktop, but the era of 10M has gradually gone. Therefore, for small and medium-sized enterprises, 10M/ 100M network card should be the first choice when purchasing.
Pay attention to the bus interface mode-the common bus interface mode in desktop and notebook computers can be found from mainstream network card manufacturers. However, it is worth noting that it is difficult to find a 100M network card with ISA interface in the market. Since 1994, PCI bus architecture has increasingly become the preferred bus for network cards, and has firmly established its position in servers and high-end desktop computers. The coming change is that this network card will be extended to all desktop computers. The high performance, ease of use and enhanced reliability of PCI Ethernet card make it widely adopted by standard Ethernet and supported by PC industry.
Network card compatibility and application technology-Fast Ethernet generally adopts 100BaseTX technology at the desktop level, and the transmission medium is UTP. Therefore, the fast Ethernet network card has RJ45 interface. Because twisted pair is widely used as transmission medium and structured wiring in small office networks, a single RJ45 interface network card can be selected. A network card with good applicability must pass the authentication of major operating systems, and at least be driven by the following operating systems: Windows, Netware, Unix, OS/2. The intelligent network card has its own processor or specially designed AISC chip, which can undertake some tasks that are undertaken by the computer processor when using non-intelligent network cards, so it rarely takes up the computer's memory and CPU time even in the case of large network information traffic. Intelligent network card has good performance and high price, and is mainly used for servers. In addition, some network cards make a fuss about BootROM to increase anti-virus function; Some network cards cooperate with the motherboard and wake up with some software. Open? LAN (Remote Wake-up) function, which can remotely start the computer through the network; Other computers just integrate the network card into the motherboard.
Network card manufacturers-Due to the maturity of network card technology, Ethernet card manufacturers are not only foreign companies such as 3Com, Intel and IBM, but also manufacturers in Taiwan Province Province have the advantages of strong production capacity and many factories in mainland China, and their prices are relatively cheap.