Detailed explanation and use of redis data types: key and string types

Keep creating , Accelerate growth ！ This is my participation 「 Nuggets day new plan · 6 Yuegengwen challenge 」 Of the 11 God , Click to see the event details

In detail Redis Of key and String The underlying principle of data types , And the most basic way to use it .

1 Redis Data type of

Redis It's not an ordinary key-value Storage server , It's actually a data structure server , Supports different types of values , stay Redis The median is not limited to simple strings , It can also contain more complex data structures .

Here are Redis All data structures supported ：

1. Binary system （Binary-safe） Safe String（ character string ）.
2. List： Sorted by insertion order String A collection of elements , Basically a linked list .
3. Set： Unique 、 Unordered String A collection of elements .
4. Sorted set： Be similar to Set, But every String Elements are associated with a floating-point value , It's called a score . Elements are always sorted by their scores , Therefore, Sets Different , It can retrieve a series of elements （ for example , Get the top ten scores or the bottom ten scores ）.
5. Hash： It consists of fields associated with values map, Fields and values are strings , This is related to Ruby or Python Of Hash Very similar .
6. Bit array( Or call it bit map)：Bitarray It's not a data structure in itself , In fact, it is String, It's just Redis Support the use of special commands can be like （bitarray） Handle string values the same as bit groups , You can set and clear individual bit position , Calculate all settings to 1 Of bit position , Find the first set or unset bit Bits, etc . This makes Redis Of String Can be used to store 、 obtain 、 Statistics and other operations , Very space saving .
7. HyperLogLog： This is a probabilistic data structure , Used to quickly estimate the cardinality of a set .
8. Stream：Redis 5.0 Version of the new data structure , For implementing persistent message queues .

2 Redis key

Redis key yes String type , It's also binary safe , We can use any binary sequence as a key , From the image “foo” Such a string goes to JPEG The content of the document .

About Redis Of key There are the following rules and recommendations ：

1. Too long key Not a good choice , Not only because of memory consumption , And look in the data key It also requires many complex key Compare .
2. Too short a key value is usually not a good choice , If you want to use ”u:1000:pwd” Instead of ”user:1000:password”, There is nothing wrong with this in itself , But the latter is easier to read , And the space consumption thus increased is relative to key object and value object Small in itself . Of course , No one is stopping you from saving a little space with shorter keys .
3. It's best to stick to a naming pattern . for example ：”object-type:id:field” It's a good idea , like this ：“user:1000:password”. For multi word field names, you can add a dot , Like this ：“comment:1234:reply.to” sample ：.
4. The biggest allowed key The size is 512 MB.

keys The command is used to return the... Of the service specified mode key list , The command is blocking , During the blocking period Redis Service not available , have access to scan command , This command extracts the specified pattern without blocking key list , But there is a certain probability of repetition , You can perform de duplication on the client .

2.1 Binary security and SDS

stay Redis The study of , Will see a lot of binary safe Description of , such as key、String wait , So what is binary security ？ It's easy to understand ： Only care about binary strings , Don't care about the specific format , Will only strictly follow the binary data access , Do not attempt to parse data in a special format , That is, how binary data is written , What happens when reading .

C The default string of the language is '\0' At the end of the , That is to say, the string you save contains '\0',c The language will recognize the previous data , The latter will be ignored , So it's not safe . such as str=“redis cluster” This string , stay C The result of direct reading in is “redis”, and strlen(str) The result of calculating the length of the string is also 4, This is not equal to the original data , So binary is unsafe .

Redis Although it is adopted C Language to develop , however Redis The string data structure stored internally is implemented by itself , It doesn't follow C Language string data structure .

redis use SDS（simple dynamic string） To implement simple dynamic strings （sds.h/sds.c）, At the same time, it guarantees redis The stored data is binary secure , The structure is as follows ：

struct sdshdr {
    int len;
    int free;
    char buf[];
};
 Copy code 

In fact, the bottom layer of the string is a char Array （ and Java Of ArrayList similar ）, At the same time, it also preserves len and free Two attributes .

len Indicates the length of the actual contents of the string ,free Represents all space allocated to the string － The actual content length of the string , That is to say free Indicates the free space length of the string , So you take the value of this string by len Property to determine the length of the actual content , Then take the value of .

String concatenation is appended to free In space . therefore redis To calculate the length and update the contents of the string C Language is much faster , Because to find the length, you only need to return the len Property value ,C The language wants to traverse the entire string to know the length . String concatenation “ commonly ” There is no need to reallocate space , The concatenated string is placed directly in free Just in memory .

Because of the definition of string length len, Therefore, when processing strings, zero value bytes are not used (\0) Mark the end of the string , Can get the complete input data , That is, binary is guaranteed .

3 Redis String

String Is the most basic Redis Value type , It's also Memcached Unique value type in .String Is the most basic Redis Value type , It's also Memcached Unique value type in . because Redis key Is string , So when we also use string type as value , We map one string to another .String The maximum length of the value is 512 MB.

Redis Of String It's binary safe , It means Redis String It can contain any type of data , for example JPEG Image or serialized Ruby object , In order to improve the running speed of the website , have access to String Type cache some static files , Such as picture file 、CSS Documents, etc. .

We can use redis-cli Order to Redis Perform command line operations . The most basic operation is SET and GET command ：

127.0.0.1:6379> set xx yyy
OK
127.0.0.1:6379> get xx
"yyy"
 Copy code 

Use SET On command , If the same has already existed key The cache of , Will perform value Replacement ：

127.0.0.1:6379> set xx yyy
OK
127.0.0.1:6379> get xx
"yyy"
 Copy code 

have access to SETNX To avoid the above situation ,SETNX yes “SET if Not eXists( If it doesn't exist , be SET)” Abbreviation . Only in key In the absence of , Will key key Is set to value . if key Already exist , be SETNX Order no action , The original value will not be overwritten .

127.0.0.1:6379> GET a
(nil)
127.0.0.1:6379> SETNX a aa
(integer) 1
127.0.0.1:6379> SETNX a bb
(integer) 0
127.0.0.1:6379> GET a
"aa"
 Copy code 

Another interesting command is that the operation is GETSET command , It is specified key Set the new value and return the original value . What's the use of this ？ for example ： Your system uses whenever new users access it INCR Command to operate a Redis key. If you want to collect this information every hour . You can start every hour GETSET This key And assign it 0 And read the original value .

127.0.0.1:6379> get num
"100"
127.0.0.1:6379> GETSET num 0
"100"
127.0.0.1:6379> get num
"0"
 Copy code 

3.1 Numerical atomic operation

Even if Redis The basic value of is a string type , You can also perform some special operations , For example, atomic increment （ The premise is that the string of integer type ）.

127.0.0.1:6379> set num 100
OK
127.0.0.1:6379> incr num
(integer) 101
127.0.0.1:6379> incr num
(integer) 102
127.0.0.1:6379> get num
"102"
 Copy code 

INCR The command parses the string value to 10 It's binary 64 Bit signed integer data , Add one , Finally, set the obtained value to the new value . There are other similar commands , Such as INCRBY—— Increase the specified value 、DECR—— Self reduction 1 and DECRBY—— Minus the specified value . In the internal , It's actually the same command , Perform... In a slightly different way .

127.0.0.1:6379> INCRBY num 20
(integer) 122
127.0.0.1:6379> decr num
(integer) 121
127.0.0.1:6379> decr num
(integer) 120
127.0.0.1:6379> DECRBY num 20
(integer) 100
 Copy code 

If key A string that is not an integer numeric type , Using the increase or decrease operation will return an exception ：

127.0.0.1:6379> set flo 11.1
OK
127.0.0.1:6379> INCR flo
(error) ERR value is not an integer or out of range
 Copy code 

If specified key non-existent , So before the increase or decrease operation , Will first set its value to 0：

127.0.0.1:6379> DECR numm
(integer) -1
 Copy code 

Even if multiple clients are connected to the same key Issue the command of increase / decrease operation , It will never lead to competitive situations . For example, the following can never happen ：“ client 1 And the client 2 Simultaneous reading 10, They both added to it 11, Then set the new value to 11”. The final value must be 12.

because INCR、INCRBY、DECR、DECRBY Such numerical operations are atomic , So it can be used to do many useful things , The most common use case is the counter .

The idea is ： Every time there are related operations , As to the Redis Server sends a incr command . For example, such a scene ： We have a web application , We want to record how many times each user visits this website every day .web The application only needs to be spliced by the user id And a string representing the time of the day key, Every time a user visits this page, it is necessary to key Execute it. incr command .

This scenario can be extended in many ways :

1. By combining INCR and EXPIRE command , It can realize a counter that only records the access times of users in the specified interval .
2. The client can use the GETSET Command gets the value of the current counter and resets it to 0.
3. Through similar DECR perhaps INCRBY Equal atomic increment / Decreasing order , Some values can be increased or decreased according to the user's operation , Like online games , The user's game score needs to be controlled in real time , Scores may increase or decrease .

In addition, for the e-commerce field , There is also an application called secsha business .

3.2 The batch operation

Redis Support for setting or retrieving multiple... In a single command key The ability to value , This is also useful for reducing latency .

Redis Provides MSET and MGET command ：

127.0.0.1:6379> MSET a aa b bb c cc d dd
OK
127.0.0.1:6379> MGET a b d
1) "aa"
2) "bb"
3) "dd"
 Copy code 

When using MGET when ,Redis Returns an array of values .

3.3 key General operation

Some commands are not specific to a particular value type , Can be associated with any type of key Use it together .

for example , Use EXISTS Command judgment key Whether the corresponding value exists , Will return 1 or 0. You can also pass in multiple key, Will return the existing key The sum of ：

127.0.0.1:6379> keys *
1) "b"
2) "c"
3) "d"
4) "a"
5) "num"
127.0.0.1:6379> EXISTS a
(integer) 1
127.0.0.1:6379> EXISTS a b
(integer) 2
127.0.0.1:6379> EXISTS a b e
(integer) 2
127.0.0.1:6379> EXISTS e
(integer) 0
 Copy code 

Use DEL The command can delete key Corresponding value , Will return 1 or 0, The identification value is successfully deleted ( The value is ) Or not deleted (key The corresponding value does not exist ). You can also pass in multiple key, Will return successfully deleted key The sum of ：

127.0.0.1:6379> DEL a c
(integer) 2
127.0.0.1:6379> DEL a c
(integer) 0
127.0.0.1:6379> DEL b
(integer) 1
127.0.0.1:6379> DEL a
(integer) 0
 Copy code 

TYPE The command can return key The storage type of the corresponding value ：

127.0.0.1:6379> KEYS *
1) "d"
2) "num"
127.0.0.1:6379> TYPE num
string
127.0.0.1:6379> TYPE d
string
127.0.0.1:6379> DEL num d
(integer) 2
127.0.0.1:6379> TYPE num
none
127.0.0.1:6379> TYPE d
none
 Copy code 

Can be applied to any value type key Set timeout , When this time comes key-value Will be deleted . Precision can be in milliseconds or seconds , But the resolution of the expiration time is always 1 millisecond .

Usually use EXPIRE To set the timeout , The default unit is seconds , You can also call this command again to change the timeout ：

127.0.0.1:6379> set a aa
OK
127.0.0.1:6379> set b bb
OK
127.0.0.1:6379> EXPIRE a 8
(integer) 1
127.0.0.1:6379> GET a
(nil)
127.0.0.1:6379> GET b
"bb"
 Copy code 

The expired information is copied and saved on disk , actually Redis Will save key Expiration time point of , When Redis When the server stops , Time has actually passed , This expires on restart key place it on clipboard .

Make in key Use before expiration PERSIST The command can remove key Timeout for ：

127.0.0.1:6379> SET a aa
OK
127.0.0.1:6379> EXPIRE a 8
(integer) 1
127.0.0.1:6379> PERSIST a
(integer) 1
127.0.0.1:6379> GET a
"aa"
 Copy code 

We can also set key-value Set the timeout when , Use TTL see key The rest of the time ：

127.0.0.1:6379> SET a aa ex 8
OK
127.0.0.1:6379> TTL a
(integer) 4
127.0.0.1:6379> TTL a
(integer) 2
127.0.0.1:6379> GET a
(nil)
 Copy code 

The following is a summary of timeout related commands ：

1. EXPIRE take key The lifetime of is set to seconds .
2. PEXPIRE： take key The lifetime of is set to milliseconds .
3. EXPIREAT： take key The expiration time for is set to timestamp The timestamp of the number of seconds represented .
4. PEXPIREAT： take key The expiration time for is set to timestamp Timestamp of the number of milliseconds represented .
5. PTTL： Check... In milliseconds key The rest of the time .

3.5 Colored eggs

Redis Of LOLWUT [VERSION version] The command will return a string containing the generated computer art image , And with Redis Version text . The demo version of this article is Redis 6.2.

LOLWUT VERSION 5：

LOLWUT：

Official documents ：www.redis.com.cn/commands/lo…

Related articles ：

1. redis.io/topics/data…
2. redis.io/topics/data…

If you need to communicate , Or the article is wrong , Please leave a message directly . In addition, I hope you will like it 、 Collection 、 Focus on , I will keep updating all kinds of Java Learning blog ！