Android interview Online Economic encyclopedia [constantly updating...]

Preface

This article is for my interview and free time to sort out the common questions of network interview . For my review . Share with you . You can comment and add ！
Because the common questions in the interview are usually the network layer , Transport layer and application layer . So I went from top to bottom , To share

Last updated time ：2021.12.30

One 、 Application layer FAQs

1.DNS How to get ip Address

Suppose the domain name is m.xyz.com Your host wants to know another host （ The domain name is y.abc.com ） Of IP Address . for example , host m.xyz.com Plan to send mail to the host y.abc.com. At this time, you must know the host y.abc.com Of IP Address . Query steps ：
① host m.xyz.com First to its local domain name server dns.xyz.com Make recursive queries .
② Local domain name server adopts iterative query . It first queries a root domain name server .
③ Root domain name server tells local domain name server , The next top-level domain name server to query dns.com Of IP Address .
④ Local domain name server to top-level domain name server dns.com The query .
⑤ Top-level domain server dns.com Tell local domain name server , The domain name server that should be queried next time dns.abc.com Of IP Address .
⑥ The local domain name server to the permission domain name server dns.abc.com The query .
⑦ Domain name server dns.abc.com Tell local domain name server , Of the host being queried IP Address .
⑧ The local domain name server finally tells the host the query result m.xyz.com
notes ：
① Used 8 individual UDP message .
②m.xyz.com The host name of is m. Top level domain name is com
③ The benefits of using iterative queries ： It can greatly reduce the load of the root domain name server , Make... On the Internet DNS The number of query requests and reply messages is greatly reduced .
Upper figure

2.http and https The difference between

① The biggest difference is https Based on the ssl above , The transmitted data is encrypted （ Combination of symmetric encryption and asymmetric encryption ）. and http No,
②https You need to apply for a certificate , Generally, there are fewer free certificates , So there will be some expenses
③https and http The ports of are respectively 443 and 80
④ Generally speaking http Page response speed ratio https fast . because HTTP Use TCP Three handshakes to establish a connection , Client and server need to exchange 3 A package , and HTTPS except TCP Three bags of , Plus ssl Handshake needs 9 A package , So the total is 12 A package .

3.http1.0、http1.1、http2.0 The difference between

（1）
HTTP 1.0 The browser and the server only maintain a short connection , Each request requires a connection with the server TCP Connect . If you need to establish a long connection , You need to set a non-standard field ,keep-alive
（2）
①HTTP1.1 in , Long connection is supported by default （Connection: keep-alive）
②HTTP 1.1 It also allows the client not to wait for the result of the last request to return , You can make the next request . But the server must send back the response results in the order of receiving the client's requests , To ensure that the client can distinguish the response content of each request
③HTTP1.1 Added more request headers and response headers . And new request methods , Such as put.
（3）
①HTTP/2.0 Support multiplexing . Reuse TCP Connect , In a connection , Both the client and the browser can send multiple requests or responses at the same time , And it doesn 't have to be 11 in sequence , This avoids ” The team's head is jammed ”
（http1.1 Although reuse is allowed TCP Connect , But the same TCP Connect the inside , All data communication is in order , The server can only process one request , Then the next request will be processed . If the front processing is particularly slow , There will be a lot of requests waiting in line .HTTP2.0 Is fully multiplexed , Not ordered and blocked .）
②HTTP2.0 Support header compression . That is, using the header table . Store previously sent key value pairs , For the same data , No longer sends per request and response .
③HTTP2.0 Support server push .

4. Enter... In the browser www.baidu.com After the implementation of the whole process

① Browser direction DNS The server sends a request to resolve the domain name ;(ip agreement 、UDP agreement ,ARP agreement ,OSPF agreement ）
②DNS The server resolves the domain name to the corresponding IP Address , And return it to the client ;
③ Browser and server setup TCP Connect ( The default port is 80,https by 443);
④ The browser sends out HTTP Request message ;
⑤ Server reply HTTP response message ;
⑥ Browser parse response message , And displayed in Web On the web ;
⑦ End of receiving and sending message , Release TCP Connect .

Two 、 Common problems of transportation layer

1.TCP How to achieve reliable transmission

① Establishing a connection ： Three handshakes
② send data ：
use window To send and receive data （ The data has serial number ）, Yes Acknowledgement mechanism , Timeout retransmission mechanism . besides , also Flow control and congestion control .
③ Connection release ： Four handshakes

2.TCP Three handshake process

Just look at the picture

Why use 3 The second handshake , instead of 2 And then ？
The triple handshake is used to prevent invalid connection request segments from suddenly being transmitted to the host B, So there's a mistake . The invalid connection request segment refers to ： host A The connection request made was not received by the host B The confirmation of , So after a while , host A Again to the host B Send a connection request , And it's built successfully , Complete data transmission in sequence . Consider such a special case , host A The first connection request sent was not lost , It's because the network node causes delays to reach the host B, host B Thought it was the mainframe A New connections started again , So the host B Agree to connect , And to the host A Send back confirmation , But at this point the host A Don't pay any attention to , host B Just waiting for the host A send data , Lead to a host B Waste of resources .

3.TCP The whole process of four waves

Just look at the picture

Be careful ：
（1） Why do you need to wave four times ？ because TCP It's full duplex , Client to server FIN When , It means that the client does not send data to the server , But the server can send data to the client , The client can also receive data .
（2） Four waves , Not necessarily four times , Middle two steps , It is possible to merge
（3） Why TIME_WAINTING state ？
① be in TIME_WAINTING In the state of , Application is not allowed to release port , etc. TIME_WAINTING Time for , The server will transmit all the data , Release again at this time
② If the client sends a breakup confirmation signal , The fourth wave , lost , Then the server will send again , If the range TIME_WAINTING Time for , Then the third wave sent by the server , The client cannot accept

4.TCP Congestion control

（1） How to judge congestion ？ let me put it another way , What will reduce the congestion window ？
① Retransmission timer timeout （ It means it's congested ）
② Received three identical ACK（ It indicates that there may be congestion ）
（2） When congestion occurs , How to control , Or how to prevent congestion ？
The way ： Slow start , Congestion avoidance , Fast retransmission , Fast recovery

5.TCP The difference between flow control and congestion control

Flow control is realized through the receiving window , Generally, the receiver , The receiver sends a confirmation message , To regulate the sending window of the sender . It's not a global .
Congestion control is realized through congestion window , Generally, the line is considered . To control the overall situation , Both the receiver and the receiver are considered , Also consider the sender , Also consider the line between the receiver and the sender . When the line is congested , The line will regulate the congestion window .
So the final sending window is The receiving window of the receiver and the congestion window of the line minimum value decision

6.TCP unpacking / What is a sticky bag

unpacking , Just more than one point , Sticky package , It's all in one . That is to say TCP Reuse and reuse
How to solve TCP Sticking and unpacking characteristics of ？
① The message was made into a fixed length
② Or the message header includes the length of the message
③ Use special delimiters （ such as ftp agreement , No messages , Use carriage return line feed , To distinguish between different messages ）

7.UDP Where to use it

① need multicast Of ,TCP Only unicast
② Packet loss does not affect the normal use of software
③ Software that requires high real-time performance

UDP Implemented software , Don't want him to lose his bag , What do I do ？
The upper layer uses itself to realize reliable transmission . Such as UDT. It is said that HTTP3 Will use UDT To achieve

8.TCP and UDP What's the difference

TCP Connection oriented , There are three handshakes and four waves ,UDP No connection , So it can also be used for broadcasting
TCP It's reliable , Specifically, timeout retransmission （ Calculate how long to wait for retransmission according to the round-trip delay ） And response confirmation
TCP There is data sorting , That is, at the opposite end , When the groups are combined , How to sort by number
TCP With flow control （ The sliding window ） And congestion control
Generally speaking ,UDP Send data faster , Real time is better than TCP stronger

9.socket What is it

The operating system hides all the details related to network communication at the bottom , use socket To operate , It's a facade pattern , call connect Method is connected .TCP use The host IP Address plus port number on the host As TCP End of connection , This kind of endpoint is called a socket .
It can be understood as the transport layer , And the following agreements , It's all encapsulated . From the perspective of application layer to the next layer , Namely socket

3、 ... and 、 Network layer FAQs

1.IP Classification of addresses

A Class address ： With 0 start , First byte range ：1~126（1.0.0.0 - 126.255.255.255）;
B Class address ： With 10 start , First byte range ：128~191（128.0.0.0 - 191.255.255.255）;
C Class address ： With 110 start , First byte range ：192~223（192.0.0.0 - 223.255.255.255）;
D Class address ： With 1110 start , First byte range ：224~239（224.0.0.0 - 239.255.255.255）;（ Use as multicast ）
All host numbers are 1 Of is the broadcast address

ip The address and subnet mask are bitwise AND , Get the network number

2.ARP How the protocol works

① First , Each host will be on its own ARP Create a... In the buffer ARP list , To indicate that IP Address and MAC Correspondence between addresses .
② When the source host wants to send data , First check ARP Is there a corresponding... In the list IP Address of the destination host MAC Address , If there is , Send data directly , without , Send... To all hosts of this network segment ARP Data packets , The data package includes ： The source host IP Address , The source host MAC Address , Of the destination host IP Address .
③ When all hosts of the network receive the ARP Packet time , First check... In the packet IP Is the address your own IP Address , If not , Then ignore the packet , If it is , First, take out the... Of the source host from the packet IP and MAC Address written to ARP In the list , If it already exists , Coverage , Then put your own MAC Address write ARP Response package , Tell the source host that it is what it is looking for MAC Address .
④ Source host received ARP After responding to the package . Will target the host's IP and MAC Address write ARP list , And use this information to send data . If the source host has not received ARP Response packets , Express ARP The query fails .
Broadcast transmission ARP request , Single broadcast to send ARP Respond to .

3. Problems related to routing protocol

RIP,OSPF,BGP Which floor do they belong to , In fact, it is still controversial . In my submission RIP,BGP It should be in the application layer ,OSPF It's better at the transport layer . because RIP use UDP,BGP use TCP,OSPF use IP. So it's divided like this

（1） Internal gateway protocol
①RIP It's the internal gateway protocol . It believes that a good route is that it passes through a small number of routers , Short distance . Not more than 15 jump .
Three characteristics
regular And the only Adjacent routers Exchange each other's All information
Update algorithm ： Drawing interpretation
②OSPF It's for customer service RIP Disadvantages of R & D . namely “ Shortest path first ”, Using the dijestra algorithm
RIP It can only be applied to small-scale networks ,OSPF It can be applied to large-scale networks . It divides an autonomous system into several smaller areas .
（2） External gateway protocol
BGP It is a routing protocol between different autonomous systems , Will choose something similar to “ Spokesman ” To exchange routing information .

Four 、 General questions

1. Equipment on each floor

The physical layer ： Repeater , A hub
Data link layer ： bridge , Switch
The network layer ： Router
Transport layer ： gateway

2. agreement

OSI layered （7 layer ）： The physical layer 、 Data link layer 、 The network layer 、 Transport layer 、 The session layer 、 The presentation layer 、 application layer .
TCP/IP layered （4 layer ）： Network interface layer 、 The layer 、 Transport layer 、 application layer .
Five layer agreement （5 layer ）： The physical layer 、 Data link layer 、 The network layer 、 Transport layer 、 application layer .

Network security related issues

1. All kinds of hijackings

DNS hijacked , It means putting DNS Parsed ip Address , Replace it with a wrong ip Address , It could be a phishing website
HTTP hijacked , It means hijacking http Request message or response message . terms of settlement ： use https

2. Flooding attack

Flooding attack is the message sent by the client for the first time , Source ip The address is false , So after the server sends the confirmation message , There has been no response , Just waiting for , Wasting resources