Interview shock 69: is TCP reliable? Why?

Compared with UDP Come on ,TCP Its main characteristics are three ： There is a connection 、 reliable 、 Data flow oriented . So-called “ There is a connection ” refer to TCP Connection management mechanism in , That is, the famous three handshakes and four waves , It's like making a phone call , Want normal communication , You must first establish a connection with the other party , That's what's called “ There is a connection ”, We will talk about the data flow oriented mechanism later , Our topic today is ：TCP How to ensure reliability ？ TCP The reason why reliability can be guaranteed , Mainly through the following 6 A means ：

1. The checksum
2. Confirm response
3. Over time retransmission
4. flow control
5. Congestion control
6. Discard duplicate data

Next , Let's take a detailed look at the specific implementation of these means .

1. The checksum

TCP The data format of the protocol is shown in the figure below ： ( picture source ： Xu Rusheng xxrs) As can be seen from the above figure “ The checksum ” Is stored in the TCP A data in the header ,TCP The sending end and receiving end of will adopt the same algorithm , According to the data sent, calculate a 16 Bit checksums , And the checksum will be sent to the receiving end together with the data . After receiving the data , A new checksum will be generated based on the received data , Then compare the new checksum with the passed checksum , If the checksums are the same , Then it means that there is no change in the data transmission process , It's a valid data , Otherwise, it is invalid data , Just give up .

Basic checksum algorithm

TCP/UDP/IP The checksum algorithm of such protocols is the same , The data flow is regarded as 16 Repeat the superposition calculation for bit integer stream . To calculate the test and , First, set the inspection and field to 0, then , For each in the valid data range 16 Bit for binary inverse code summation , The results are in the tests and fields , If the data length is odd, fill in one byte 0. When you get the data , Also for each in the valid data range 16 The sum of the binary inverse code of the digits . Because the receiver includes the verification and the , therefore , If there is no error in the transmission of the first part , Then the receiver's calculation result should be all 0 Or all 1( To be specific, we have realized , The essence is the same ) . If the result is not all 0 Or all 1, That means the data is wrong .

2. Confirm response

Confirmation and response mechanism is the key means to ensure the reliability of message delivery , It is also almost all message oriented middleware （MQ） in , One of the most commonly used technologies , For example, mainstream message oriented middleware RabbitMQ、Kafka、RocketMQ There is a confirmation and response mechanism in , That's what we often say ACK（ACKnowledge Character, Confirmation character ）. The confirmation response mechanism is TCP in , The core mechanism to ensure the reliability of messages . How can you confirm that the other party must have received your message ？ Undoubtedly, the most effective way is for the other party to tell you , It has received , This is the confirmation response . The process of confirmation and response is shown in the following figure ：

3. Over time retransmission

There may be two problems during the confirmation of the message ： First of all , The message was lost when it was sent , second , The message was lost when confirming the reply , As shown in the figure below ： obviously , Even if there is a confirmation and response mechanism, it can not guarantee that the message will not be lost , Then what shall I do? ？ It doesn't matter if the message is lost , After the sender confirms the loss of the message , It is not enough to compensate the receiver for the same message ？ This is the retransmission timeout mechanism .

Clever timeout retransmission mechanism

TCP The timeout retransmission mechanism of is also very clever in design , It is to ensure that messages are in any environment , Can communicate efficiently , therefore TCP It's using “ Dynamic time ” Timeout retransmission mechanism of . For example, if the message is lost for the first time , Then the sender will be 500ms Then send a message , If the second message sent is also lost , Then the sender will be 1000ms Then send a message , If the third message is also lost , Then it will 2000ms Then send a message , If you accumulate a certain number of times , The message has not been successfully sent , that TCP You will think that the other host is abnormal , Will force the connection to close , This is it. TCP The main execution process of overtime retransmission .

4. flow control

The speed of data processing at the receiving end is limited , If the sender sends too fast , This will cause the buffer at the receiving end to be full , At this time, if the sender continues to send , It will cause packet loss , Then it causes a series of chain reactions such as packet loss and retransmission . therefore TCP According to the processing of the receiving end , Dynamically adjust the size of the sent data , This mechanism is called flow control （Flow Control）.

5. Congestion control

Congestion control refers to TCP According to the current network situation , Dynamically control the amount of data sent , Transfer data at a suitable speed . Imagine , If TCP In an environment that is not clear about the network , Rashly send a large amount of data to the receiving end , This will lead to more packet loss and overtime retransmission , Thus causing a series of chain reactions , Cause data transmission to slow down . and TCP What is taken is “ Slow start ” Mechanism , Start with a small amount of data , Explore the way , Find out the current state of network congestion , And then decide how fast to transmit data , This is the congestion control mechanism . If more data is transferred , There are a lot of packet losses , that TCP It will reduce the amount of data sent , Then try to slowly increase the amount of data sent , Through this form of dynamically sending data packets , To achieve data transmission suitable for the current network speed , This is it. TCP The concrete realization of congestion control .

6. Discard duplicate data

Through the previous knowledge, we know , When confirming the response , Due to the loss of acknowledgement message , Then the receiver may receive duplicate data from the sender , As shown in the figure below ： At this time, for the business side , Only one data is needed , therefore TCP There is also a mechanism , Mechanism for discarding duplicate data , This will ensure that the data received by the business party is correct . TCP A number will be added to each package sent , If a packet with the same number is received , Then it means that the receiving end gets duplicate packets , Just throw it away .

summary

TCP The main means to ensure reliability are 6 individual ： The checksum 、 Confirm response 、 Over time retransmission 、 flow control 、 Congestion control 、 Discard duplicate data . Among them, flow control and congestion control are easy to confuse , We need to know clearly , Flow control is a control mechanism for the receiving capacity of the receiving end , Congestion control is a control mechanism for the current network , So don't mix it up .

Reference resources & Acknowledgement

blog.csdn.net/namelcx/article/details/6866720

It's up to you to judge right and wrong , Disdain is to listen to people , Gain or loss is more important than number .

official account ：Java Analysis of the real interview questions

Interview collection ：https://gitee.com/mydb/interview