一文详解MySQL实现分库分表的详细步骤

-- 先看看现在的“房子”有多大
SELECT 
    TABLE_SCHEMA as '数据库',
    TABLE_NAME as '表名',
    ROUND((DATA_LENGTH + INDEX_LENGTH) / 1024 / 1024, 2) as '大小(MB)',
    TABLE_ROWS as '行数'
FROM information_schema.TABLES 
WHERE TABLE_SCHEMA = '你的数据库名'
ORDER BY (DATA_LENGTH + INDEX_LENGTH) DESC;

-- 思考人生三连问：
-- 1. 按业务分？(用户相关、订单相关、商品相关)
-- 2. 按时间分？(2024订单、2025订单)
-- 3. 按地域分？(北京用户、上海用户)

-- 原来都挤在一个库里
CREATE DATABASE single_apartment;

-- 现在买栋楼，每层一个专业户
CREATE DATABASE user_villa;      -- 用户专属楼层
CREATE DATABASE order_mansion;   -- 订单豪华层
CREATE DATABASE product_tower;   -- 商品展示层

-- 用户表搬到用户楼层
CREATE TABLE user_villa.user_info (
    user_id BIGINT PRIMARY KEY AUTO_INCREMENT,
    username VARCHAR(50) NOT NULL,
    email VARCHAR(100),
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
    -- 用户相关的其他字段...
);

-- 订单表搬到订单楼层  
CREATE TABLE order_mansion.order_info (
    order_id BIGINT PRIMARY KEY AUTO_INCREMENT,
    user_id BIGINT NOT NULL,
    amount DECIMAL(10, 2),
    status TINYINT,
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
    -- 订单相关的其他字段...
);

-- 订单表按月份拆分，再也不怕找去年的订单像考古了
-- 2024年订单表
CREATE TABLE order_mansion.order_202401 (
    order_id BIGINT PRIMARY KEY AUTO_INCREMENT,
    user_id BIGINT NOT NULL,
    -- ... 其他字段
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    INDEX idx_user_id (user_id),
    INDEX idx_created_at (created_at)
) PARTITION BY RANGE (MONTH(created_at)) (
    PARTITION p1 VALUES LESS THAN (2),
    PARTITION p2 VALUES LESS THAN (3),
    -- ... 更多分区
);

-- 2024年2月订单表
CREATE TABLE order_mansion.order_202402 (
    -- 结构同上
);

-- 用户表按ID取模分表，16张表够不够？
-- 先创建用户表模板
DELIMITER $$

CREATE PROCEDURE create_user_tables()
BEGIN
    DECLARE i INT DEFAULT 0;
    WHILE i < 16 DO
        SET @table_name = CONCAT('user_villa.user_info_', LPAD(i, 2, '0'));
        SET @sql = CONCAT('
            CREATE TABLE IF NOT EXISTS ', @table_name, ' (
                user_id BIGINT PRIMARY KEY AUTO_INCREMENT,
                username VARCHAR(50) NOT NULL,
                email VARCHAR(100),
                created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
                INDEX idx_username (username)
            ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4
        ');
        PREPARE stmt FROM @sql;
        EXECUTE stmt;
        DEALLOCATE PREPARE stmt;
        SET i = i + 1;
    END WHILE;
END$$

DELIMITER ;

-- 执行创建
CALL create_user_tables();

// Java代码示例：数据路由导航
public class ShardingNavigator {
    
    // 用户表路由：根据user_id决定去哪个表
    public static String getUserTableName(long userId) {
        int tableIndex = (int) (userId % 16);
        return String.format("user_info_%02d", tableIndex);
    }
    
    // 订单表路由：根据时间决定去哪个表
    public static String getOrderTableName(Date orderTime) {
        SimpleDateFormat sdf = new SimpleDateFormat("yyyyMM");
        return "order_" + sdf.format(orderTime);
    }
    
    // 组合查询：跨表查询就像组织家庭聚会
    public List<Order> getUserOrders(long userId, Date startTime, Date endTime) {
        List<Order> allOrders = new ArrayList<>();
        Calendar calendar = Calendar.getInstance();
        calendar.setTime(startTime);
        
        // 遍历时间范围内的所有月份表
        while (calendar.getTime().before(endTime)) {
            String tableName = getOrderTableName(calendar.getTime());
            String sql = "SELECT * FROM " + tableName + " WHERE user_id = ?";
            // 执行查询并添加到结果集
            // ...
            calendar.add(Calendar.MONTH, 1);
        }
        return allOrders;
    }
}

# ShardingSphere配置文件示例（YAML格式）
# 这位管家知道所有房间在哪
dataSources:
  ds0:
    url: jdbc:mysql://localhost:3306/user_villa?useSSL=false
    username: root
    password: your_password
  ds1:
    url: jdbc:mysql://localhost:3307/order_mansion?useSSL=false
    username: root
    password: your_password

shardingRule:
  tables:
    user_info:
      actualDataNodes: ds0.user_info_${0..15}
      tableStrategy:
        inline:
          shardingColumn: user_id
          algorithmExpression: user_info_${user_id % 16}
    order_info:
      actualDataNodes: ds1.order_${2024..2025}${1..12}
      tableStrategy:
        standard:
          shardingColumn: created_at
          preciseAlgorithmClassName: com.example.TimeShardingAlgorithm
  
  bindingTables:
    - user_info,order_info

-- 第一步：先抄家（备份）
CREATE TABLE order_mansion.order_202401_new LIKE order_mansion.order_202401;

-- 第二步：慢慢搬（增量迁移）
INSERT INTO order_mansion.order_202401_new
SELECT * FROM single_apartment.orders 
WHERE created_at >= '2024-01-01' 
  AND created_at < '2024-02-01';

-- 第三步：检查有没有落下的
SELECT 
    (SELECT COUNT(*) FROM single_apartment.orders 
     WHERE created_at >= '2024-01-01' AND created_at < '2024-02-01') as old_count,
    (SELECT COUNT(*) FROM order_mansion.order_202401_new) as new_count;

-- 第四步：切换门牌号（重命名）
RENAME TABLE 
    order_mansion.order_202401 TO order_mansion.order_202401_backup,
    order_mansion.order_202401_new TO order_mansion.order_202401;

// 雪花算法：Twitter出品，必属精品
public class SnowflakeIdGenerator {
    private final long twepoch = 1288834974657L;
    private final long workerIdBits = 5L;
    private final long datacenterIdBits = 5L;
    private final long sequenceBits = 12L;
    
    private long workerId;
    private long datacenterId;
    private long sequence = 0L;
    private long lastTimestamp = -1L;
    
    public synchronized long nextId() {
        long timestamp = timeGen();
        
        if (timestamp < lastTimestamp) {
            throw new RuntimeException("时间倒流了，检查系统时间！");
        }
        
        if (lastTimestamp == timestamp) {
            sequence = (sequence + 1) & 4095;  // 2^12-1
            if (sequence == 0) {
                timestamp = tilNextMillis(lastTimestamp);
            }
        } else {
            sequence = 0L;
        }
        
        lastTimestamp = timestamp;
        
        return ((timestamp - twepoch) << 22) |
               (datacenterId << 17) |
               (workerId << 12) |
               sequence;
    }
    
    // 全局唯一，趋势递增，适合分库分表
}

// 使用Seata等分布式事务解决方案
@GlobalTransactional
public void transferMoney(long fromUser, long toUser, BigDecimal amount) {
    // 1. 从用户库扣款
    // 2. 向订单库插入记录
    // 要么都成功，要么都回滚
}

-- 分表前：一步到位
SELECT * FROM orders WHERE user_id = 123;

-- 分表后：变成侦查行动
SELECT * FROM order_202401 WHERE user_id = 123
UNION ALL
SELECT * FROM order_202402 WHERE user_id = 123
UNION ALL
-- ... 查遍所有月份表

// 1. 先从用户库查用户
List<User> users = userDao.getUsers(condition);
List<Long> userIds = users.stream().map(User::getId).collect(Collectors.toList());

// 2. 再去订单库查这些用户的订单
List<Order> orders = orderDao.getOrdersByUserIds(userIds);

// 3. 在内存中组装
Map<Long, List<Order>> userOrdersMap = groupOrdersByUser(orders);