TiDB 表分区

前置条件

对比MySQL的表分区

使用之前，要先读官方文档

---

---

---

表分区，注意事项

1. 确认表中是否存在唯一键(包括主键) ，如果存在，那么唯一键 中，必须包含分区表达式中用到的所有列
2. 如果表中不存在唯一键 ，那么第1条规则也就不需要遵守了
3. 分区列是主键列，那么分区列不可以是 NULL
4. 分区表达式中列的数据类型，要与表中列的数据类型相符合
5. 删除分区一定会删除分区中所有数据，不会自动转移数据
6. MAXVALUE 因为添加分区必须是递增的，添加此分区时要注意，此分区表示已经是最大的分区，无法在添加新的分区。

---

---

---

1 表分区 RANGE

-- 创建表
CREATE TABLE table_range (
  a bigint(20) DEFAULT NULL,
  b varchar(255) DEFAULT NULL,
  c int(11) DEFAULT NULL,
  d date DEFAULT NULL,
  KEY m_index (a,b,c)
)
-- 创建 RANGE 表分区 按照年分区
PARTITION BY RANGE(YEAR(d)) (
    PARTITION p0 VALUES LESS THAN (1900),
    PARTITION p1 VALUES LESS THAN (2010),
    PARTITION p2 VALUES LESS THAN MAXVALUE
);

-- 添加数据
INSERT INTO table_range VALUES (1,'1',1,'1900-01-01'), (2,'2',2,'2009-02-02'), (3,'3',3,'2019-03-03');

-- 添加表分区
-- ALTER TABLE table_range ADD PARTITION (PARTITION p3 VALUES LESS THAN (2014));

-- 清空表分区
-- ALTER TABLE table_range TRUNCATE PARTITION p1;

-- 删除表分区
-- ALTER TABLE table_range DROP PARTITION p3;

-- 查询数据 1900~2009 之间的数据，这里只会查询 p0 表分区
SELECT * FROM table_range WHERE d > 1900 AND d < 2009;

-- 指定查询 p1 表分区
SELECT * FROM table_range PARTITION (p1);

---

-- 创建表
CREATE TABLE table_range (
  a bigint(20) DEFAULT NULL,
  b varchar(255) DEFAULT NULL,
  c int(11) DEFAULT NULL,
  d date DEFAULT NULL,
  KEY m_index (a,b,c)
)
-- 创建 RANGE 表分区 按照 年、月、日 分区
PARTITION BY RANGE(TO_DAYS(d)) (
    PARTITION p0 VALUES LESS THAN (TO_DAYS('2017-02-01')),
    PARTITION p1 VALUES LESS THAN (TO_DAYS('2017-03-01')),
    PARTITION p2 VALUES LESS THAN MAXVALUE
);

---

-- 创建表
CREATE TABLE table_range (
  a bigint(20) DEFAULT NULL,
  b varchar(255) DEFAULT NULL,
  c int(11) DEFAULT NULL,
  d TIMESTAMP DEFAULT NULL,
  KEY m_index (a,b,c)
)
-- 创建 RANGE 表分区 按照 时间戳 分区
PARTITION BY RANGE(UNIX_TIMESTAMP(d)) (
    PARTITION p0 VALUES LESS THAN (UNIX_TIMESTAMP('2017-02-01 00:00:00')),
    PARTITION p1 VALUES LESS THAN (UNIX_TIMESTAMP('2017-03-01 00:00:00')),
    PARTITION p2 VALUES LESS THAN MAXVALUE
);

---

查询系统表中记录的表分区

SELECT
    PARTITION_NAME AS '分区名称',
    FROM_DAYS(PARTITION_DESCRIPTION) AS '原分区条件',
    PARTITION_EXPRESSION AS '分区表达式',
    PARTITION_DESCRIPTION AS '分区条件',
    TABLE_ROWS AS '分区中数据行数'
FROM
    INFORMATION_SCHEMA.PARTITIONS
WHERE
    TABLE_SCHEMA = 'prd2_pfizer'
    AND TABLE_NAME = 'dc_flowdata_deliver_sale_range';

---

---

---

2 表分区 HASH

-- 创建表
CREATE TABLE table_hash (
  a bigint(20) DEFAULT NULL,
  b varchar(255) DEFAULT NULL,
  c int(11) DEFAULT NULL,
  d date DEFAULT NULL,
  KEY m_index (a,b,c)
)
-- 创建 HASH 表分区
PARTITION BY HASH(YEAR(d))
PARTITIONS 6;

-- 添加数据
INSERT INTO table_hash VALUES (1,'1',1,'1900-01-01'), (2,'2',2,'2009-02-02'), (3,'3',3,'2019-03-03');

案例

CREATE TABLE `dc_flowdata` (
  `distributor_code` varchar(200) COLLATE utf8_bin DEFAULT NULL COMMENT '客户编码',
  `distributor_name` varchar(200) COLLATE utf8_bin DEFAULT NULL COMMENT '客户名称',
  `specification` varchar(200) COLLATE utf8_bin DEFAULT NULL COMMENT '产品规格',
  `manufacturer` varchar(200) COLLATE utf8_bin DEFAULT NULL COMMENT '生产厂家',
  `lot_no` varchar(200) COLLATE utf8_bin DEFAULT NULL COMMENT '产品批号',
  `expiry_date` date DEFAULT NULL COMMENT '有效期',
  `paas_create_ntid` varchar(32) COLLATE utf8_general_ci DEFAULT '' COMMENT '创建人NTID',
  `paas_update_ntid` varchar(32) COLLATE utf8_general_ci DEFAULT '' COMMENT '修改人NTID',
	
  PRIMARY KEY (`biz_date`,`paas_id`),
  KEY `index1` (biz_date,paas_is_del,normal_product_code,normal_organ_code,br_status,distributor_code,normal_inventory_state),
	KEY `index_source_file_id` (`source_file_id`)
	
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci COMMENT='库存数据'



PARTITION BY RANGE(TO_DAYS(biz_date)) (
    PARTITION p200001 VALUES LESS THAN ( TO_DAYS('2000-01-01 00:00:00') ),
	PARTITION p201701 VALUES LESS THAN ( TO_DAYS('2017-02-01 00:00:00') ),
	PARTITION p201702 VALUES LESS THAN ( TO_DAYS('2017-03-01 00:00:00') ),
	PARTITION p201703 VALUES LESS THAN ( TO_DAYS('2017-04-01 00:00:00') ),
	PARTITION p201704 VALUES LESS THAN ( TO_DAYS('2017-05-01 00:00:00') ),
	PARTITION p201705 VALUES LESS THAN ( TO_DAYS('2017-06-01 00:00:00') ),
	PARTITION p201706 VALUES LESS THAN ( TO_DAYS('2017-07-01 00:00:00') )

);

只有以下函数可以用于分区表达式：

ABS()
CEILING() (see CEILING() and FLOOR())
DATEDIFF()
DAY()
DAYOFMONTH()
DAYOFWEEK()
DAYOFYEAR()
EXTRACT() (see EXTRACT() function with WEEK specifier)
FLOOR() (see CEILING() and FLOOR())
HOUR()
MICROSECOND()
MINUTE()
MOD()
MONTH()
QUARTER()
SECOND()
TIME_TO_SEC()
TO_DAYS()
TO_SECONDS()
UNIX_TIMESTAMP() (with TIMESTAMP columns)
WEEKDAY()
YEAR()
YEARWEEK()