In computer network engineering, we have always heard of several layers, some say seven layers, some say five layers. In fact, even if we don’t know what each layer is, we should have heard of the application layer, physical layer and so on. Here is a detailed look at how these so-called layers of computer networks came to be and what they represent.

 

One, about OSI model

1. What is OSI model?

OSI stands for Open System Interconnection Reference Model, which translates as OSI and Speaks as a network Interconnection Model. ISO (International Organization for Standardization) is a conceptual model published in 1984.

We call it the OSI reference model, but note that it is a model that describes abstract concepts that computer vendors should consider. No matter how it is implemented, if everyone is following the model, then everyone can communicate with each other.

 

2. How many layers does the OSI reference model have?

The OSI reference model has seven layers, which is arguably the most comprehensive, given that other models have fewer layers.

From lowest to highest:

Physical layer, data link layer, network layer, transmission layer, session layer, presentation layer, application layer

Each layer implements its own functions and protocols, and completes interface communication with adjacent layers.

Since it is a layer structure, of course the lower layer is the foundation of the upper layer, which provides more advanced services based on the support of the lower layer. Just like a building, each floor is closely related to each other.

 

3. Explain the role of each layer

Here we outline what each layer is used for, which is more detailed than other articles, but still far from enough. Because each layer has a lot of content, I’m just going to write a little summary here.

Start with the low to the high

① Physical layer (PH)

The physical layer is the most important and basic layer in the OSI layered architecture. It is the network oriented layer and involves the protocols related to the physical part of the network.

The physical layer uses some physical transmission media, such as twisted pair, coaxial cable, optical fiber, and wireless channel, to transmit information. The task of the physical layer is to provide a physical connection for the upper layer, as well as the mechanical, electrical, functional and process characteristics expressed by the physical connection, to achieve transparent bitstream transmission.

At this level, the data is not yet organized. The sender only sends the signal, the receiver only detects the signal, it does not care how many bits make up each data unit, nor does it care about the meaning of the data being transmitted, only as raw bit stream submitted to the upper layer (data link layer). Finally, the physical layer is responsible for the physical topology or actual network layout. The physical layer includes the network cabling, hub, repeater, and NIC.

Personal Summary:

Anyway, the physical layer is to open up the path and transmission of data, no matter how abstract and complex the upper layer, ultimately have to rely on it to transmit data, and generally transmitted are binary signals (bitstream). We don’t care what device it is, wired or wireless, because it itself is trying to shield physical devices and transmission media from the next level of detail. In this way, the upper layer is not affected, no matter how the physical layer transmits data wirelessly, it does not affect the functions and protocols of the upper layer.

The physical plane is like a delivery robot that has no idea what it is transporting. It can take a car, a plane, or a boat and just deliver the package to the other side. The next layer, the data link layer, is like an employer and doesn’t need to know how it delivered the package, just whether it arrived or not.

 

② Data-link Layer (D)

The data link layer is the second layer in OSI, located between the physical layer and the network layer. Its main function is to realize reliable data transmission on unreliable physical lines. The data unit of this layer is Frame, which is composed of three parts: Frame header, data part and Frame tail. Among them, the frame header and tail contain some necessary control information, such as synchronization information, address information (MAC address), error control information, etc., while the data part contains the data transmitted from the network layer.

If transmitted over Ethernet, the data frame can be called Ethernet data frame or Ethernet frame. The structure of an Ethernet frame looks like this from beginning to end:

  • Presync code: Used to enable the receiving adapter to quickly adjust the clock frequency to match that of the sender when receiving a MAC frame. It can be understood as preheating before formal transmission of frame data.
  • Frame start: The start of a frame, which is used to tell the receiving adapter that frame information is coming.
  • Destination address: When receiving a frame, the receiver checks the address to see if it is destined for itself, and if not, discards it.
  • Source address: The physical address (MAC address) of the network adapter that sent the frame. This tells you who sent it to you.
  • Type: indicates the type of the upper-layer protocol. Because there are too many protocols at the upper-layer network layer, this field must be set when data is processed. For example, if the field is 0x0800, data is delivered to the IP protocol.
  • Data: Also known as payload, represents the data delivered to the upper layer. The minimum length of a frame is 46 bytes and the maximum length is 1500 bytes. If it is less than 46 bytes, it is filled to the minimum length. The maximum value is also called the maximum transmission unit (MTU).
  • Frame check code (FCS) : Detect whether the frame is wrong. This part can be understood by yourself. It is mainly to detect whether the data is lost or changed in the transmission process. Error control.
  • Frame interval: the end of a frame after the end of the frame check code. However, an interval must be set before the next frame, and at least 12 OCTET idle line status codes must be sent.

About the data link layer error control and flow control, here is not to say, want to know more about it you can go to understand. Error control is mainly judged by some codes, such as parity check code, cyclic redundancy code CRC and Hamming code. The flow control is mainly to control the sending rate of the sender, to ensure that the receiver can not handle it, specific stop-wait protocol, back N frame protocol, select retransmission protocol, etc.

Personal Summary:

The physical layer helps us transmit signals, but how does it know where to go? Is that correct? This requires the data link layer, which encapsulates the data passed down from the previous layer into something called a data frame, and then passes it to the physical layer as bits.

In fact, this data frame is before and after the data with the frame header and the frame tail. In the frame header, there are some start symbols, their address, the address of the target, the type of the upper protocol information, through which we can locate the place to be transmitted to ensure that the data is transmitted to the correct location. At the end of the frame is the check code for error control. By ensuring the correctness of data transmission, the faulty data frame needs to be retransmitted.

The addresses at the data link layer are MAC addresses, or NIC addresses. Each NIC has a unique address

You can see that the data link layer also does not process the data itself, it just takes the data from the upper layer and wraps it up, and at the same time supervises the physical layer to ensure the correctness of the data transmission.

It works like a delivery system for express delivery, confirming the last stop and the next stop (to a transfer station), and they don’t need to know much about what’s inside, just what kind. When the package arrives at the next station, check to see if it arrived correctly and if it is missing or broken. It also doesn’t care how the Courier delivers the goods, it just needs to control the right delivery.

 

③ Network Layer (N)

The network layer is the highest layer of the communication subnet and the interface between the communication subnet and the resource subnet. The network layer mainly serves the upper layer (transport layer), it still does not process the data, it just helps the transport layer transfer the data. It is responsible for addressing and planning the best transport path, and maintaining routing tables. Determine which route is the fastest and most appropriate, and when to use an alternative router. It also has the ability of congestion control and flow control.

The data unit sent by the network layer is called Packet. It is similar to the previous data frame and is also a message container dedicated to this layer, but it can contain much more data than the data link layer. Take IP Packet of IP protocol as an example, its maximum length is 64K (65535) bytes. Even in Ethernet transmission, the maximum MTU(Maximum transmission unit) is only 1500. In this case, data fragmentation is required. A packet is divided into dozens of pieces, and if a link with a smaller MTU is encountered during transmission, data fragmentation is required again.

Take the IP packet as an example and take a quick look at its structure:

  • Version: indicates the IP version number of the packet. In the command, 0100 indicates IP version 4,0110 indicates IP version 6, which is usually IPv4.

  • Header length: The length of the packet header. It indicates how many 32-bit long integers the packet header contains. That is, how many 4-bytes of data.

  • Service type: Consists of eight bits, each representing some package configuration such as importance, delay, traffic, reliability, cost, etc.

  • Package length: Indicates the total length of the entire packet, in bytes. In this case, 64K (65535 bytes).

  • Reassembly identifier: Used for fragment reassembly. Fragments of the same IP packet use the same identifier.

  • Flag: Configuration for sharding, including three bits, but one bit is reserved, so there are two configurations: whether sharding is allowed, and whether it is the last packet.

  • Slice offset: The offset used for the fragment, in bytes, indicating the relative position of the current fragment in the original IP packet. Generally, the first shard is 0, and the second shard needs to calculate the offset based on the data length of the previous shard.

  • TTL (Time To Live) : This is the TTL seen in the ping command. Although it looks like a Time concept, it is actually the maximum number of hops that an IP packet can be forwarded in a computer network. Each Time an IP packet is forwarded, its number decreases by 1. Its maximum value is 255 and the recommended value is 64.

  • Protocol code: indicates the upper-layer protocol that uses the package, for example, TCP=6, ICMP=1, UDP=17.

  • Header check and: used to check the integrity and correctness of IP packets’ headers.

  • Source Start IP address and destination IP address: indicates the source IP address and destination IP address.

  • Optional: Some special cases may be used.

  • User data: Data information for packages.

 

Personal Summary:

The network layer is often referred to as the IP layer because this layer has the core IP protocol that establishes a connection between two terminals through IP addressing. A Packet is also called an IP Packet because of IP. But the protocol for IP is not defined in OSI, because OSI is only responsible for abstraction, while IP protocol is already a concrete implementation class.

The data link layer is mainly connected to other terminals through MAC, which means it can only be found when the two terminals are connected. Once it passes through the network, it is impossible to find the data link layer by MAC. At this time need network, subnet, IP those blah blah blah, network layer is stem this, even if the distance is very far, as long as there is a path in the middle, it can pass the 99 81 difficult to get to the West.

At the same time, it also has the function of error control and sequence control. As we know, there is error control in the data link layer to ensure the complete and correct transmission of each data frame, so the error control in the network layer is only aimed at the head of the packet, it only checks the packet head. Just make sure that the data headers are fine (but of course the network layer doesn’t have to control the data again because there is another layer of error control at the transport layer).

The sequence control is because the distance is long, we may not every packet is the same path, and there may be packet loss and retransmission in the middle, so that the order of arrival will be different. Sequence control is only about sorting packets in order. It does not force packets to be received in order. The network layer in this layer is still a worker, is to help the upper people work.

The network layer, for example, is just like the receiving station of express delivery. When receiving a parcel, it identifies the sender and recipient, and then plans the best route to which transfer station. If the package is too big for the next level of a single delivery, it needs to be divided into several packages and of course assembled and delivered to the next level when it is received.

 

④ Transport Layer (T)

The first three layers are mainly responsible for data communication, that is, the “tool people” who transmit data. The transmission layer is just in the middle of the seventh layer, which also plays a role of connecting the preceding and the following. It provides end-to-end data exchange mechanism, error control and flow control, and provides reliable transmission service for three layers, namely session layer, presentation layer and application layer.

There are two important protocols at this layer: TCP and UDP. TCP is connection-oriented and provides reliable transport, while UDP is connectionless and unreliable.

There are a lot of differences, but here’s a little bit. TCP protocol we are more familiar with some, every time it is to establish a connection before transferring data, after the transfer release connection, and an error control, flow control, timer, repeatedly confirmed, etc., including four times we heard three times handshake, waving it in order to guarantee the security and stability of the data transmission is reliable, has made a lot of effort. The overhead of natural transmission needs to be higher. The most commonly used application-layer protocols, HTTP and FTP, are based on TCP (although you may not know what they are, you must have heard of them).

And UDP as said earlier, it is not reliable, because it is a connectionless, so want to pass the data, no matter what each other, no matter it does not exist, even want to send it, can be said to be very follow one’s inclinations, also because of this, its transmission efficiency is higher than the TCP will be many, but also because of this, it’s not security is too high, it is easy to lost package or error. So if you want to use UDP, be sure to use other means to control and ensure that the message tolerance rate.

UDP is packet-oriented. If the upper application layer gives it a packet as long as it wants, it will keep the packet as it is and then add the head to send it to the network layer. It is almost lazy to do any processing on it and does not worry about whether the packet is too big or too small. TCP is byte stream oriented, although the upper layer and it is also an interaction of a size of data block, but in the eyes of TCP, this is just a series of unstructured byte stream, it will receive the byte stream processing, select the appropriate size for transmission.

Data units at this layer are called data segments or packet segments. Use TCP as an example.

  • Source port and destination port: two bytes each, ranging from 0 to 65535.

  • Data sequence number: 4 bytes. TCP is byte stream oriented. Each byte has a sequence number, which stores the sequence number of the first byte of the current datagram. For example, if the current packet has a serial number of 301 and carries 100 bytes of data, the sequence number of the next packet is 401.

  • Acknowledgement sequence number: 4 bytes. The acknowledgement sequence number is the sequence number of the next message, which tells the receiver to expect it to receive next time. These two serial numbers ensure that the SEQUENCE of TCP packets is correct.

  • Offset: 4 bits. It indicates the distance between the start of the TCP packet and the start of the TCP packet. It also indicates the length of the TCP header, because the length of the TCP header is not fixed.

  • Reserved word: 6 bits, reserved bit, temporarily unused, should be set to zero.

Then there are 6 bits, which are mostly used in shaking hands and waving inside, you can go to understand the combination.

  • URGent bit URG(URGent) : When URG is 1, the URGent pointer field is valid. It tells the system that there is urgent data in this message segment and that it should be transmitted as soon as possible (equivalent to high-priority data).

  • ACKnowledgment bit ACK(ACKnowledgment) : The ACKnowledgment number field is valid only if ACK = 1. When ACK = 0, the confirmation number is invalid. It is equivalent to enabling the confirmation number function.

  • PuSH bit PSH(PuSH) : After a process on one end sends a packet, it needs to receive the response immediately. You can set PSH=1 so that the receiver delivers the message to the receiving application as soon as it is received (that is, “pushed” forward), rather than waiting until the entire cache fills up and delivering it up.

  • Reset bit RST: When RST is 1, it indicates that a serious error occurs in the TCP connection (for example, the host crashes or other reasons) and the connection must be released before the transport connection is re-established.

  • SYNchronization bit SYN(SYN) : SYNchronization bit If SYN is set to 1, it indicates that this is a connection request or connection acceptance packet. Is the first packet sent in a TCP handshake.

  • Terminate bit FIN(FINal) : Used to release a connection. If the FIN value is 1, the data on the sender end of the packet segment is sent and the transport connection is released.

  • Window field: 2 bytes. The window field is used to control the amount of data sent by the other party and changes dynamically. Because the data cache space of the receiver is limited, the window number is required to let the sender know how much data the receiver is allowed to send.

  • Packet checksum: 2 bytes, including header and data checksum. When calculating the checksum, a 12-byte dummy header is added to the front of the TCP packet segment.

  • Emergency pointer: 2 bytes. The emergency pointer is meaningful only when URG=1 and indicates the sequence number of the last byte of emergency data in this paragraph.

  • Optional options and fill bits: Optional options are additional configurations that can be up to 40 bytes long, while the fill bits are used to make up the numbers, because if the optional options do not make up a multiple of 4 bytes (32 bits), the fill bits need to be padded.

  • User data: Formal data information.

As you can see, the TCP header is quite complex, which is necessary to establish reliable communication channels.

 

Next comes UDP, whose data units, called datagrams, are much simpler than TCP’s. The header is only about 8 bytes, whereas TCP’s header is at least 20 bytes, which is one of the reasons it’s fast. Its header reads as follows:

  • Source port: the 16-bit source port number. This port is used when the reply from the peer party is needed, and 0 is used when no reply is required.
  • Destination port: the 16-bit destination port number must be used when the destination delivers packets.
  • Length: The length of the header and data in bytes. 16 bits. The minimum value is 8 because the header is at least 8 bytes.
  • Checksum: 16-bit checksum used to check whether UDP datagrams have errors in transmission and discard any errors.

UDP has a very simple header, which is as short as possible.

 

Personal Summary:

The third layer is only responsible for data regardless of transmission, the third layer is only responsible for transmission regardless of data, and the transport layer in the middle is to do both sides. At the same time, the concept of port is also very important, is only in the network layer IP to IP, and at the transport layer to is port to port, we usually start the program, visit the web site, set up agencies, these operations in addition to need IP port, only specific to the port, can the accurate transmission of data to the computer in a process.

At the same time, it also carried out the last error control and flow control. After all, the data link layer in front is a short line connection, which can only manage the transmission of two nodes. And the transmission layer is the overall planning, the overall data error control and flow control. It is because of its overall control that it can be called reliable. The transmission layer is the last level of data transmission, to the higher level there is no need to take care of data transmission.

TCP and UDP are the core of the transport layer. One is connected and the other is connectionless. Although we think TCP is the god forever at first glance, UDP has so many problems, what is the use of fast? But in fact UDP fast is also meaningful, such as QQ is largely based on UDP, but did not see what problems it is not? Because QQ based on UDP to ensure the efficiency of data transmission, but also use other means to ensure the accuracy of the message and error tolerance rate. There are live streaming apps, online voice programs, and so on. You can’t send and receive signals every time you say a word, can you? So UDP also has its function.

After receiving the data, the transport layer selects the connectionless or connectionless protocol (TCP/UDP) based on the upper-layer protocol type, and then encapsulates it into data segments or datagrams with unique headers. Then is whether to establish a connection, if there is no connection, directly thrown to the network layer transmission is done, do not have to consider so much. If there is a connection, it is necessary to have a few back to play tai chi, until the confirmation of the communication line and both ends of the transceiver data function is no problem, to formally establish the connection, and then also need to verify each message is accurate transmission. That’s what the transport layer does, and it takes a lot more work to be connected than unconnected, because it’s reliable.

 

⑤ Session layer (Session, S)

Finally at the top, but there’s not as much to say.

The session layer is built on top of the transport layer and is responsible for establishing, managing, and terminating sessions between two entities, including authentication and permission authentication. Of course, it does not establish the connection itself, but gives it to the transport layer to do, and this session and transport layer connection is different, although also based on it, but compared to more abstract, virtual. It is also responsible for addressing, finding the location of the target, such as IP and port by domain name. This layer also has error checking, but it no longer deals with errors in the data transfer process, but with more advanced errors such as disk space, lack of paper in the printer, etc.

Take SMB protocol as an example, it is a protocol of Windows system to share files in LAN. Although in Baidu Encyclopedia, it should be a protocol of session layer and presentation layer as well as a small part of application layer, it does not prevent us from illustrating some functions of session layer with it.

When host A shares A folder with host B, host B needs to access [\(host A’s IP address)] to access the shared folder of HOST A. This is A session request. However, when you log in to host A for the first time, the system prompts you to enter host A’s account and password. Otherwise, the system cannot access host A. Otherwise, host C will be denied access without authorization. This is the authentication and authentication link of the session layer.

When the session is successfully established, B can copy A’s file to the local machine, but when B is finished, after the sharing window is closed, the transfer must be over, and the transport layer is unlikely to continue to waste time there. But when we access the shared address again, we find that we do not need to enter a password, but can still access. This is the session layer holding the session, and since it is holding the session rather than a transport layer connection, the session is not actually finished.

The session will be closed only after configuration, such as closing the window to close the session, or automatically closing the session after an hour. Then, as at the beginning, you will need to enter the account password again because the session has been closed and needs to be re-established.

In addition, in the session layer and beyond, data is collectively referred to as messages, and there is no special name for them, because they are no longer concerned with transmission.

Personal Summary:

The session layer, in turn, serves the presentation layer and is only responsible for initiating and managing the session, but the transport layer does all the real dirty work, which is capitalist. And don’t compare it to our Web development sessions, which, while they sound similar, aren’t really the same thing.

And to be honest, the session layer and presentation layer can be said very little, because they are almost never successfully implemented alone, are implemented together with the application layer of the three layers. In the TCP/IP model, these three layers are simply consolidated into one application layer. OSI is ultimately an abstract model, but there is no place for the session or presentation layer in its current use.

 

⑥ Presentation layer (Presentation, P)

The sixth presentation layer is responsible for data processing. Control the syntax and format of data, encode and transform data, compress and decompress data, and encrypt and decrypt data. The proper processing of data through this layer ensures compatibility, even between different operating systems.

Personal Summary:

Once you get to the presentation layer, it has nothing to do with data transfer and connection at all, it just handles data, or represents data. For example, two machines with different coding formats will have completely different meanings if they use different coding to parse the same sentence. Images, audio, and video data also need to be converted before they can be transmitted or returned to the application layer. This is all done by the presentation layer, which is the data-only layer. In practice, of course, it is rarely isolated, and in the TCP/IP model, it is also in the application layer.

 

⑦ Application Layer (Application, A)

The final layer is the application layer, which also has the most protocols. Application-layer protocols define rules for communication and interaction between processes specific to specific applications. Different applications also need different application-layer protocols, so the number of protocols is so large.

Common protocols such as file Transfer Service (FTP), domain name System (DNS), hypertext Transfer Protocol (HTTP), SMTP protocol support email, and so on.

Personal Summary:

It’s the last layer, and it doesn’t have anything to do with the bottom layer anymore, it’s kind of a user-facing layer. In the TCP/IP model, it even merges the presentation layer and the session layer directly, so the number of protocols is predictably large and miscellaneous.

However, the application layer is ultimately a layer, not a concrete application. It is an abstraction in OSI and a model in TCP/IP models, which are not specific to the concrete application. Although we have almost touched it, there is a barrier in the middle. For example, when we open a web page, we are directly exposed to the browser (Google or Internet Explorer), and the browser is based on the application layer protocol HTTP. We use FTP to download files, probably using some FTP client software installed, but this software must be developed based on FTP. Or when we send an email, it may be Outlook, or it may be web pages open online, but they must be based on SMTP and POP3 protocol.

This is true of the application layer, which is not application-specific, but serves the application.

 

End summary layer

I added this layer, of course, but actually OSI is only 7 layers. When I was studying network engineering, I had heard the OSI 7-layer protocol for countless times, but I still couldn’t remember it completely. Now that we understand the meaning of each layer and their functions, I found it was not difficult to write it down.

From the very bottom of the physical layer, it’s like a dedicated porter, just doing the lifting, no matter what.

Then, the data link layer is mainly responsible for controlling the physical layer, because the physical layer only transmits and is not reliable, while the data link layer is to make the unreliable line reliable through logical means (error control, flow control).

The network layer manages data transmission from a higher perspective, and it is also the highest layer of the communication subnet. At present, location is mainly carried out through IP, which may pass through many nodes and data link layer transmission for many times.

The next layer is the transport layer, which is not personally involved in any transmission process, but is in charge of the data transmission, providing an end-to-end data exchange mechanism. In addition, there are two modes of connection and no connection, if there is a connection of TCP, the existence of this layer is very important, completely rely on it to establish a reliable connection, if it is UDP, its role will be relatively small, equivalent to just do a layer of data encapsulation and simple processing to the network layer.

Then there is the high-level session layer, which controls the session between the two entities and also performs authentication and authentication. Here it does not participate in any specific data transfer, but only controls the session, making a virtual continuous connection.

Then there is the presentation layer, which is specialized in data processing, format conversion, encryption and decryption, compression and decompression, it will be processed all kinds of data will be handed over to the next layer.

Finally is the highest levels of application layer, session layer and presentation layer can be said to be its wings, although the OSI ideal is good, very specific points, but in practical use, the three layers are merged into an application layer, because they are so inseparable, different application layer the presentation layer and session layer difference is too big, it is difficult to separate out as a layer, So it’s good to know a little bit about the top two layers. The application layer mainly serves applications. It has many typical protocols, such as HTTP, FTP, DNS, etc. It is the layer with the most protocols.

 

 

Second, TCP/IP architecture

OSI, now we’re going to study the TCP/IP architecture, and that’s going to explain the love-kill story between the two models, but let’s take a quick look at what TCP/IP is.

1. What is TCP/IP

TCP/IP (Transmission Control Protocol/Internet Protocol) is a Protocol cluster that transfers information between multiple networks. TCP/IP is not only TCP and IP, but a protocol cluster consisting of FTP, SMTP, TCP, UDP, and IP. It is called TCP/IP because TCP and IP are the most representative of TCP/IP.

So, you see, TCP/IP is usually not just TCP and IP, but their entire protocol family, a bunch of protocols. The TCP/IP hierarchy has four layers: application layer, transmission layer, network layer and network interface layer.

In fact, we can probably know the meaning of these four layers, because it can also be said to refer to OSI stratification, which we have illustrated a lot in the previous example.

2. Differences between OSI and TCP/IP

As we know, OSI is the Open System Interconnection Model, which is a model of rules designed to allow different computers to communicate with each other. As long as according to its model to achieve, can realize the communication between different systems (on the network interconnection). The model is very detailed and thoughtful, with seven layers, each with a different set of protocols. And OSI is the ISO organization launched, to know that ISO is the authority, it makes things, are world-class standards, according to what it says to do the right.

The next step would have been to wait for major software/hardware vendors to implement it, but it didn’t catch on because the standards were too complex and “perfect” to be easy to implement. There was also a TCP/IP protocol that basically implemented network communication and was much simpler than OSI, so why not use an existing one?

So it can be understood that OSI lost to TCP/IP in terms of practicality, of course OSI is not completely obsolete, because its layered theory and ideas are still very worthy of reference (probably because it is only used as reference, so it is now called reference model).

TCP/IP’s four-tier model can also be seen as a reference to OSI, as shown in figure:

The application layer, presentation layer, and session layer of three layers are merged into the application layer, reducing unnecessary layers. The physical layer and data link layer are merged into the network interface layer.

We’re not going to go through each layer, because we’ve already done a lot, but we just need to know what each layer does.

 

3. Where did the five-tier architecture come from?

Now we know OSI has seven layers and TCP/IP model has four, but what we often see on the network is a five-layer architecture. Where is that architecture?

In fact, it is just a combination of the two models, because in this way, it is not very complicated, but also can be explained clearly in a comprehensive way. The specific levels are as follows:

Physical layer, data link layer, network layer, transmission layer, application layer

Basically, it’s the third floor.

 

 

Third, summary

This article focuses on the OSI reference model in conjunction with the TCP/IP model. Although we may not know the details, we can at least get a sense of what the seven layers are all about.

Among them, TCP, UDP and IP protocols are the areas that need to be studied, but in view of the space, we will leave them to be explained separately next time.

 

 

 

OSI Model — Baidu Baike

Baike.baidu.com/item/OSI%E6…

Diagram of Ethernet data frame format

C.biancheng.net/view/6391.h…

TCP/IP packet structure details

Blog.csdn.net/prsniper/ar…

TCP, UDP header details

Blog.csdn.net/weixin_4273…

OSI seven-layer model and TCP/IP five-layer model

www.cnblogs.com/qishui/p/54…