?????????????????????????????????????????? TCP / IP protocol
??????????????????????? http://91mail.51.net is available for learning exchange
?????????????????????? To make more translation documents please http://91mail.51.net download ??????????? ????????????????????????????????????? First part IP: The target of the network protocol network is to provide a dense Communication Systems. To achieve this goal, the network protocol software must hide the details of some network itself, and provide convenient conditions for a large number of virtual networks. The operation of the virtual network is the same as any other network, allowing computers to accept and send packets. The main difference between the virtual network and the real network is that the virtual network is only an abstract imagination of the designer and is completely created by software. Designers can freely select address package formats and can transmit some hardware technology details independently. ?? The address is a key component of the network. To give an independent, unified system's appearance, all hosts must be configured with a unified address. Unfortunately, the address of the actual network is not enough because a internet can contain multiple network technologies and each technology defines its own network format. In this way, the address used by the two techniques may be incompatible because they have different models or formats. ?? In order to ensure that all hosts have a unified address, the protocol software defines a settlement scheme independent of the actual location. Although the Internet addressing scheme is an abstract concept established by the software, the agreement address is used as a destination file for a virtual network, similar to the hardware as a destination file of the actual network. In order to send a packet on the Internet, the sender is set to the destination protocol address in the packet and transmitted through the protocol software. This software uses the protocol address of the destination when transmitting the packet through the Internet to the destination computer. ?? Since the detail of hidden potential networks, unified addressing helps generate a large number of inexplicable networks. Two applications communicate with each other without knowing the other party's hardware address. This idea I have done makes some users aware that the agreement address is provided by software and is not part of a computer system. Interestingly, we will learn about some level of software protocols also use the agreement address. In the Internet, in order to provide a unified addressing mode, the agreement software defines an abstract address scheme that assigns a unique location of each host. Users, applications, and high-level software protocols communicate with abstract addresses. In the TCP / IP Agreement Stack, addressing is specified by the Internet Agreement. IP standard specifies that each host is assigned a special 32-bit Internet protocol address, usually abbreviated as IP address, Or ID. Each packet sent by the Internet is included with the IP address of the sender and the 32 bit of the recipient. Therefore, through TCP / IP transmission information, a computer must know the IP address of the remote computer to be sent. • Therefore, a network address is a 32-bit binary number for a host, which is used to communicate with the host. Conceptually, each 32-bit IP address is distinguished into two parties: prefix and suffix; two-layer structure design is to make the router take effect. The prefix indicates the actual network connected to the computer, and the suffix indicates that the Personal computer works above. That is, each actual network is assigned in the Internet with a unique network number. The network number is expressed in an address prefix in the connected computer. In addition, the computer on each given actual network is assigned a unique location suffix. Although there is no two networks can be assigned the same network number, and there is no two computers on a network can be assigned the same suffix, but a suffix value can not only be used on one network. For example, if an internet contains three networks, they may be assigned the network number 1, 2, 3. The three computers of the network may be assigned a suffix of 1, 3, and 5, and three computers that depend on the network 2 can be assigned to 1, 2, and 3. The hierarchy of the IP address ensures two important performance: (1) Each computer is assigned a unique address. (2) Although the allocation of network numbers is a certain global coordination, there is no global reconciliation that can be distributed locally. So the first characteristic guarantees a complete address that contains prefixes and suffixes to ensure uniqueness. If the two computers are attached to different actual networks, their location has a different suffix.
Once they choose to select size for the IP site and decide to divide each address into two parts, IP designers must determine how many parts are. The prefix requires a full number of bits to make each actual network on the Internet to a unique network number. The suffix requires sufficient number of bits to connect to each computer on the network to the unique suffix. It is impossible to have a simple choice, because give a part of the added bit means to reduce the number of digits from elsewhere. Choosing a large prefix adapts to many networks but limits the size of each network; selecting a large suffix means that each actual network can contain many computers, but limits the total number of networks. • Because an Internet can include any network technology, an Internet can construct from some big actual networks, while additional possibilities may consist of many small networks. More importantly, a single Internet can contain large and small networks. As a result, the designer selects a solution that can adapt to large and small networks. The solution divides the IP memory space into five, A, B, C, D, E, each with different sizes of prefixes and suffixes. ?? Because A, B, and C are used to be host address, they are called the main layer. The D layer is used as a broadcast layer that allows information to be transmitted to a set of computers. In order to use IP multi-broadcast, a group of hosts must allow a multi-targeted order. Once the multipoint broadcast group is established, the copy of any packet sent to multiple broadcast addresses will be sent to each host in the form of a group. ?? Just as a number display, the main layer is divided into a prefix and suffix by bytes. The A layer is divided into the first and second bytes. The B layer divides the second and third bytes. The C layer divides the third and fourth bytes. Summary, IP divide the host address into three main layers. Each layer of the address determines the boundaries of the network prefix and the host suffix. TCP: Transmission Control Protocol ?? Program designer recognizes reliability is the basis of a computer system. For example, a programmer may be requested to establish an application to send data to the I / O device such as the printer. The application writes data to the device, but does not need to confirm whether the data is completely free. Instead, the application depends on the basic computer system to ensure a reliable transmission; the system guarantees that the data will not be lost, replicated, or the order is chaotic. When creating an application with an Internet, in order to let the programmer follow traditional technologies, the software on the Internet must provide the same language as a traditional computer system. That is to say, the software must ensure fast, reliable communications. Data must be sent under completely identical order, and must not have lost or copies. Reliability is the responsibility of the transfer agreement; the application is used to send and accept data with a transfer service interaction. In the Transmission Control Protocol / IP program group, the transfer control protocol provides a reliable transmission service. TCP is very important because he can solve the problem very well, although other agreements have been created, but other transfer protocols have not been found better than him. Therefore, most of the Internet applications are based on TCP. Therefore, the reliability provided by the transfer agreement is the basis of other applications. Transmission Control Agreement (Transmission Control Protocol) is a reliable transport layer in the transfer control protocol group. From the point of view of the application, the services provided by the transfer control protocol have seven major features: (1) Based on the connection service. An application must first request a connection to the destination and then transfer data with the connection. (2) Communication of point-to-point points. Each transfer control protocol must have two points. (3) It is completely reliable. Transfer Control Agreement guarantees that a connection will be delivered completely, no data loss or confusion. (4) Full duplex communication. A transmission control protocol connection allows data bidirectional transmission, and allows any application to send data at any time. The transmission control protocol can double-way buffering and access data, which can transmit data for an application and continue to calculate during data transfer. (5) String interface. We say that the transfer control protocol provides a string interface where one of the applications can transmit records, and it is not necessary to ensure that the data is delivered to the same size by the same size block through the transmitter. (6) Reliable connection start. Transfer Control Protocol Requirements When both applications generate connections, both must agree to new connections; copies used in previous connections will not produce the correct response or contact with new connection. ?????????????????????? (7) a beautiful connection closure.
An application can open a connection, send any more data, and request to close the connection. Transmission control protocols ensure reliable delivery of all data before closing the connection. Therefore, the transmission control protocol provides completely reliable (no data copies or loss), the connection-based, full duplex water flow transfer service, allows two applications to form a connection, two partitions transmitted data, and then terminate the connection. Each transfer control protocol connection is reliably started and the end of the excellence, so that all data is transmitted before termination. • Because it provides a connection to an application directly from a computer to a remote computer, the transfer control protocol is also referred to as end-to-end protocol. Applications can request a connection, accept, and send data in the form of transmission control protocol, and turn off the connection. Because they are implemented by software, the connection provided by the transfer control protocol is called a virtual connection. In the end, exchange information between the TCP software modules on the two machines to express the connection between them. Transmission control protocol uses IP carrying information. Each transmission control protocol information is compressed in an IP packet and transmitted over the Internet. When the packet arrives on the host in the destination, the IP transmits content to the transfer control protocol. Note Although the transmission control protocol uses IP carrying information, IP is not read or interpreted. Therefore, the transmission control protocol view IP is a packet communication system that connects the host in an end-to-end connection, and IP regards each transmission control protocol information as being transferred. Part B Related to some annotations of TCP / IP From the viewpoint of implementation, the huge complex modules and agreements given will not be surprised to discover the original implementation method is extremely large, difficult to handle, slow. Everyone trying to be very anxious. It doesn't want to link "OST" and "Poor Quality". When the product becomes better during the run, it is difficult to encounter difficulties. In contrast, one of TCP / IP's first implementation is part of Berkeley UNIX and quite good. People began to use it quickly, leading to a large user community, resulting in advance, leading to an even larger community. Here the spiral rise is replaced. Because of the implementation, many people in the academic community regard TC / IP as a part of UNIX, and UNIX is not a parent-child relationship in the academic community in the 1980s (and then incorrectly known as the mother Different from apple pie. OSI, on the other hand, it is wanted to be a creativity of the Communication Department in Europe, the European Community, and the US government in the future. Belief is only part of the truth, but a group of government bureaucrats forced low-level technologies, for poor researchers and programmers have no big help in developing computers. Some people are in the future language as IBM in the 1960s, and DODD is later declared that this is actually Adam. Although the fact is that OSI models and agreements have certain success, there is still a lot of interested organizations, most European PTTs still have monopoly on communication. As a result, a weak effort was used to update OSI, published a revision model in 1994. About what is changed and what should be changed. TCP / IP models and agreements also have their problems. ? First, the model cannot be clear about the concept of the regional service, interfaces and agreements. Good software engineering practice requires some things that are very careful in specifications and implementation, and TCP / IP is not. As a result, the TCP / IP model is not a wide range of guides with new technologies. Second, the TCP / IP model is not comprehensive and difficult to describe any agreement relaxation other than itself. For example, the TCP / IP describes that SNA is almost impossible. Third, the host - the network layer is not a real in normal feeling in the layer protocol. It is an interface. (Between networks and data chains) A interface between an interface and a layer is important, it must not be sloppy. Fourth, TCP / IP model does not distinguish between physics and data chains. These are completely different. The physical layer is related to the transmission characteristics of copper, fiber and radio communication. The main task of the data chain layer is to delineate the boundaries of starting congenial and ending, and transmitted from one side from one side from one side. One suitable model should be included. The TCP / IP model did nothing.
