Stop using UUID indiscriminately. Have you tested the performance gap between self incrementing ID and UUID?

Click on the official account , Practical technical articles Know in time

Catalog

• One 、 Preparation form & data

• Two 、500w Level data test

  • 2.1 entry 500W data , Self increasing ID Save half the disk space

  • 2.2 Single data is indexed , Self increasing id and uuid Little difference

  • 2.3 Range like Inquire about , Self increasing ID Better performance than the UUID

  • 2.4 Write test , Self increasing ID yes UUID Of 4 times

  • 2.5、 Backup and recovery , Self increasing ID Better performance than the UUID

  • 500W summary

  • 1000W summary

  • Self increasing ID Primary key + step , Suitable for medium-sized distributed scenarios

  • UUID, Suitable for small-scale distributed environment

One 、 Preparation form & data

UC_USER, Self increasing ID Primary key , The table structure is similar to the following ：

CREATE TABLE `UC_USER` (
  `ID` bigint(20) NOT NULL AUTO_INCREMENT COMMENT ' Primary key ',
  `USER_NAME` varchar(100) DEFAULT NULL COMMENT ' user name ',
  `USER_PWD` varchar(200) DEFAULT NULL COMMENT ' password ',
  `BIRTHDAY` datetime DEFAULT NULL COMMENT ' Birthday ',
  `NAME` varchar(200) DEFAULT NULL COMMENT ' full name ',
  `USER_ICON` varchar(500) DEFAULT NULL COMMENT ' Head picture ',
  `SEX` char(1) DEFAULT NULL COMMENT ' Gender , 1: male ,2: Woman ,3： A secret ',
  `NICKNAME` varchar(200) DEFAULT NULL COMMENT ' nickname ',
  `STAT` varchar(10) DEFAULT NULL COMMENT ' User state ,01: normal ,02: frozen ',
  `USER_MALL` bigint(20) DEFAULT NULL COMMENT ' Current owner MALL',
  `LAST_LOGIN_DATE` datetime DEFAULT NULL COMMENT ' Last login time ',
  `LAST_LOGIN_IP` varchar(100) DEFAULT NULL COMMENT ' Finally log in IP',
  `SRC_OPEN_USER_ID` bigint(20) DEFAULT NULL COMMENT ' Federated login of source ',
  `EMAIL` varchar(200) DEFAULT NULL COMMENT ' mailbox ',
  `MOBILE` varchar(50) DEFAULT NULL COMMENT ' mobile phone ',
  `IS_DEL` char(1) DEFAULT '0' COMMENT ' Whether or not to delete ',
  `IS_EMAIL_CONFIRMED` char(1) DEFAULT '0' COMMENT ' Whether to bind the mailbox ',
  `IS_PHONE_CONFIRMED` char(1) DEFAULT '0' COMMENT ' Whether to bind the mobile phone ',
  `CREATER` bigint(20) DEFAULT NULL COMMENT ' founder ',
  `CREATE_DATE` datetime DEFAULT CURRENT_TIMESTAMP COMMENT ' Registration time ',
  `UPDATE_DATE` datetime DEFAULT CURRENT_TIMESTAMP COMMENT ' modification date ',
  `PWD_INTENSITY` char(1) DEFAULT NULL COMMENT ' Password strength ',
  `MOBILE_TGC` char(64) DEFAULT NULL COMMENT ' Mobile login ID ',
  `MAC` char(64) DEFAULT NULL COMMENT 'mac Address ',
  `SOURCE` char(1) DEFAULT '0' COMMENT '1:WEB,2:IOS,3:ANDROID,4:WIFI,5: Management system , 0: Unknown ',
  `ACTIVATE` char(1) DEFAULT '1' COMMENT ' Activate ,1： Activate ,0： not active ',
  `ACTIVATE_TYPE` char(1) DEFAULT '0' COMMENT ' Activation type ,0： Automatically ,1： Manual ',
  PRIMARY KEY (`ID`),
  UNIQUE KEY `USER_NAME` (`USER_NAME`),
  KEY `MOBILE` (`MOBILE`),
  KEY `IDX_MOBILE_TGC` (`MOBILE_TGC`,`ID`),
  KEY `IDX_EMAIL` (`EMAIL`,`ID`),
  KEY `IDX_CREATE_DATE` (`CREATE_DATE`,`ID`),
  KEY `IDX_UPDATE_DATE` (`UPDATE_DATE`)
) ENGINE=InnoDB AUTO_INCREMENT=7122681 DEFAULT CHARSET=utf8 COMMENT=' User table '

UC_USER_PK_VARCHAR surface , character string ID Primary key , use uuid：

CREATE TABLE `UC_USER_PK_VARCHAR_1` (
  `ID` varchar(36) CHARACTER SET utf8mb4 NOT NULL DEFAULT '0' COMMENT ' Primary key ',
  `USER_NAME` varchar(100) DEFAULT NULL COMMENT ' user name ',
  `USER_PWD` varchar(200) DEFAULT NULL COMMENT ' password ',
  `BIRTHDAY` datetime DEFAULT NULL COMMENT ' Birthday ',
  `NAME` varchar(200) DEFAULT NULL COMMENT ' full name ',
  `USER_ICON` varchar(500) DEFAULT NULL COMMENT ' Head picture ',
  `SEX` char(1) DEFAULT NULL COMMENT ' Gender , 1: male ,2: Woman ,3： A secret ',
  `NICKNAME` varchar(200) DEFAULT NULL COMMENT ' nickname ',
  `STAT` varchar(10) DEFAULT NULL COMMENT ' User state ,01: normal ,02: frozen ',
  `USER_MALL` bigint(20) DEFAULT NULL COMMENT ' Current owner MALL',
  `LAST_LOGIN_DATE` datetime DEFAULT NULL COMMENT ' Last login time ',
  `LAST_LOGIN_IP` varchar(100) DEFAULT NULL COMMENT ' Finally log in IP',
  `SRC_OPEN_USER_ID` bigint(20) DEFAULT NULL COMMENT ' Federated login of source ',
  `EMAIL` varchar(200) DEFAULT NULL COMMENT ' mailbox ',
  `MOBILE` varchar(50) DEFAULT NULL COMMENT ' mobile phone ',
  `IS_DEL` char(1) DEFAULT '0' COMMENT ' Whether or not to delete ',
  `IS_EMAIL_CONFIRMED` char(1) DEFAULT '0' COMMENT ' Whether to bind the mailbox ',
  `IS_PHONE_CONFIRMED` char(1) DEFAULT '0' COMMENT ' Whether to bind the mobile phone ',
  `CREATER` bigint(20) DEFAULT NULL COMMENT ' founder ',
  `CREATE_DATE` datetime DEFAULT CURRENT_TIMESTAMP COMMENT ' Registration time ',
  `UPDATE_DATE` datetime DEFAULT CURRENT_TIMESTAMP COMMENT ' modification date ',
  `PWD_INTENSITY` char(1) DEFAULT NULL COMMENT ' Password strength ',
  `MOBILE_TGC` char(64) DEFAULT NULL COMMENT ' Mobile login ID ',
  `MAC` char(64) DEFAULT NULL COMMENT 'mac Address ',
  `SOURCE` char(1) DEFAULT '0' COMMENT '1:WEB,2:IOS,3:ANDROID,4:WIFI,5: Management system , 0: Unknown ',
  `ACTIVATE` char(1) DEFAULT '1' COMMENT ' Activate ,1： Activate ,0： not active ',
  `ACTIVATE_TYPE` char(1) DEFAULT '0' COMMENT ' Activation type ,0： Automatically ,1： Manual ',
  PRIMARY KEY (`ID`),
  UNIQUE KEY `USER_NAME` (`USER_NAME`),
  KEY `MOBILE` (`MOBILE`),
  KEY `IDX_MOBILE_TGC` (`MOBILE_TGC`,`ID`),
  KEY `IDX_EMAIL` (`EMAIL`,`ID`),
  KEY `IDX_CREATE_DATE` (`CREATE_DATE`,`ID`),
  KEY `IDX_UPDATE_DATE` (`UPDATE_DATE`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COMMENT=' User table ';

Two 、500w Level data test

#  Self increasing id Table for primary key 

mysql> select count(1) from UC_USER;

+----------+

| count(1) |

+----------+

|  5720112 |

+----------+

1 row in set (0.00 sec)

 

# uuid Table for primary key 

mysql> select count(1) from UC_USER_PK_VARCHAR_1;

+----------+

| count(1) |

+----------+

|  5720112 |

+----------+

1 row in set (1.91 sec)

2.1 entry 500W data , Self increasing ID Save half the disk space

In terms of the space capacity occupied , Self increasing ID Than UUID Less than half .

2.2 Single data is indexed , Self increasing id and uuid Little difference

2.3 Range like Inquire about , Self increasing ID Better performance than the UUID

2.4 Write test , Self increasing ID yes UUID Of 4 times

2.5、 Backup and recovery , Self increasing ID Better performance than the UUID

500W summary

stay 500W Under the test of the record sheet ：

（1） Ordinary single or 20 Search about records ,uuid For the primary key, the difference is small, and the efficiency is almost the same ;

（2） But the range query, especially the query of hundreds of records , Self increasing id Is more efficient than uuid;

（3） During range query and statistical summary , Self increasing id Is more efficient than uuid;

（4） On top of the storage , Self increasing id The storage space occupied is uuid Of 1/2;

（5） On backup recovery , Self increasing ID The primary key is slightly better than UUID.

1000W summary

stay 1000W Under the test of the record sheet ：

（1） Ordinary single or 20 Search about records , The self increasing primary key efficiency is uuid Primary key 2 To 3 times ;

（2） But the range query, especially the query of hundreds of records , Self increasing id Is more efficient than uuid;

（3） During range query and statistical summary , Self increasing id The efficiency of primary keys is uuid Primary key 1.5 To 2 times ;

（4） On top of the storage , Self increasing id The storage space occupied is uuid Of 1/2;

（5） On the write , Self increasing ID The efficiency of primary keys is UUID Primary key 3 To 10 times , The difference is obvious , especially update Data in a small range .

（6） On backup recovery , Self increasing ID The primary key is slightly better than UUID.

Self increasing ID Primary key + step , Suitable for medium-sized distributed scenarios

In each cluster node group master above , Set up （auto_increment_increment）, Stagger the starting points of each cluster 1, The step size selection is greater than the number of cut diversity groups that are basically impossible to achieve in the future , Reach will ID Relative segmentation effect to meet Globally unique The effect of .

• Advantage is ： Implement a simple , Later maintenance is simple , Transparent to applications .

• The disadvantage is that ： The first setting is relatively complex , Because we need to calculate enough step length for the future business development ;

UUID, Suitable for small-scale distributed environment

about InnoDB For engines that aggregate primary key types , The data is sorted by primary key , because UUID The disorder of ,InnoDB It's going to produce huge IO pressure , And because indexes and data are stored together , String as primary key will double the storage space .

When storing and retrieving ,innodb The primary key will be physically sorted , This is right auto_increment_int It's good news , Because the primary key position of the last insertion is always at the end .

But yes. uuid Come on , This is bad news , because uuid It's messy , The primary key position of each insert is uncertain , Maybe at the beginning , Or in the middle , During physical sorting of primary keys , It is bound to cause a lot of IO Operation affects efficiency , When the amount of data keeps growing , Especially when there are tens of millions of records , The reading and writing performance is declining very severely .

• advantage ： It's easy to build , There is no need to handle the uniqueness of the primary key .

• shortcoming ： Take up twice the storage space （ It takes more money to store one piece in the cloud 2 Times the money ）, In the later stage, the reading and writing performance declines sharply .

source ：blog.csdn.net/qq_30108237/article/details/106856051

recommend

Java Interview questions

Technical involution group , Learn together ！！

PS： Because the official account platform changed the push rules. , If you don't want to miss the content , Remember to click after reading “ Looking at ”, Add one “ Star standard ”, In this way, each new article push will appear in your subscription list for the first time . spot “ Looking at ” Support us ！