作者： lmdb 原文来源： https://tidb.net/blog/3b586d36
平凯数据库在11月27日正式发布了 v.8.5.4版本，提到新特性“支持表级别数据打散功能”，虽然现在是实验特性不建议在生产库中实践，这个特性的引入可能解决日常运维中痛点问题 ， 因 TiKV 节点上的数据分布不均，造成数据热块问题。 这里做尝鲜熟悉下新特性，仅此而已，也希望这特性在下个版本中正式发布。支持表级别数据打散功能（实验特性） #63260 @ bufferflies
PD 会自动调度数据，将整个集群的数据尽可能均匀地分布到所有 TiKV 节点上。然而，这种自动调度是基于集群全局的。在某些场景下，尽管整个集群的数据分布是均衡的，但某张表在各个 TiKV 节点上的数据分布仍然可能不均匀。
从 v8.5.4 开始，你可以通过
SHOW TABLE DISTRIBUTION语句查看某张表在集群中所有 TiKV 节点上的数据分布情况。如果存在数据分布不均衡，可以通过DISTRIBUTE TABLE语句对该表进行数据打散（实验特性），以提升负载均衡性。表级数据打散功能属于一次性执行任务，并设有超时时间限制。如果到达超时时间后，打散任务还未完成，则会自动退出。
这里提到了新查看命令SHOW TABLE DISTRIBUTION和DISTRIBUTE TABLE，分别对表进行查看数据分别和打散数据，这与我们经常使用SHOW TABLE REGIONS语句显示 TiDB 中某个表的 Region 信息还是不同。
SHOW TABLE REGIONS会返回如下列：
REGION_ID：Region 的 ID。START_KEY：Region 的 Start key。END_KEY：Region 的 End key。LEADER_ID：Region 的 Leader ID。LEADER_STORE_ID：Region leader 所在的 store (TiKV) ID。PEERS：Region 所有副本的 ID。SCATTERING：Region 是否正在调度中。1 表示正在调度。WRITTEN_BYTES：估算的 Region 在 1 个心跳周期内的写入数据量大小，单位是 byte。READ_BYTES：估算的 Region 在 1 个心跳周期内的读数据量大小，单位是 byte。APPROXIMATE_SIZE(MB)：估算的 Region 的数据量大小，单位是 MB。APPROXIMATE_KEYS：估算的 Region 内 Key 的个数。SCHEDULING_CONSTRAINTS：Region 所在的表或者分区所关联的 放置策略 的规则。SCHEDULING_STATE：关联了放置策略的 Region 的当前调度状态。
SHOW TABLE DISTRIBUTION会返回如下列：
- PARTITION_NAME: STORE_ID: STORE_TYPE: REGION_LEADER_COUNT: REGION_PEER_COUNT: REGION_WRITE_BYTES: REGION_WRITE_KEYS: REGION_WRITE_QUERY: REGION_LEADER_READ_BYTES: REGION_LEADER_READ_KEYS: REGION_LEADER_READ_QUERY: REGION_PEER_READ_BYTES: REGION_PEER_READ_KEYS: REGION_PEER_READ_QUERY:
对新特性做简单的测试
MySQL [sakila]> MySQL [sakila]> CREATE TABLE tt ( -> id INT NOT NULL PRIMARY KEY auto_increment, -> b INT NOT NULL, -> pad1 VARBINARY(1024), -> pad2 VARBINARY(1024), -> pad3 VARBINARY(1024) -> )ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin -> PARTITION BY RANGE (`id`) -> (PARTITION `p1` VALUES LESS THAN (10000), -> PARTITION `p2` VALUES LESS THAN (20000), -> PARTITION `p3` VALUES LESS THAN (30000), -> PARTITION `p4` VALUES LESS THAN (40000), -> PARTITION `p-max` VALUES LESS THAN (MAXVALUE)); Query OK, 0 rows affected (0.61 sec) MySQL [sakila]> MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM dual; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 1 row affected (0.04 sec) Records: 1 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 1 row affected (0.04 sec) Records: 1 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 8 rows affected (0.01 sec) Records: 8 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 1000 rows affected (0.02 sec) Records: 1000 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 10000 rows affected (0.14 sec) Records: 10000 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 10000 rows affected (0.16 sec) Records: 10000 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 10000 rows affected (0.23 sec) Records: 10000 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 10000 rows affected (0.18 sec) Records: 10000 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 10000 rows affected (0.35 sec) Records: 10000 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 10000 rows affected (0.32 sec) Records: 10000 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 10000 rows affected (0.46 sec) Records: 10000 Duplicates: 0 Warnings: 0 MySQL [sakila]> INSERT INTO tt SELECT NULL, FLOOR(RAND()*1000), RANDOM_BYTES(1024), RANDOM_BYTES(1024), RANDOM_BYTES(1024) FROM tt a JOIN tt b JOIN tt c LIMIT 10000; Query OK, 10000 rows affected (0.36 sec) Records: 10000 Duplicates: 0 Warnings: 0 MySQL [sakila]> select count(*) from tt; +----------+ | count(*) | +----------+ | 81010 | +----------+ 1 row in set (0.11 sec) MySQL [sakila]> SELECT SLEEP(5); +----------+ | SLEEP(5) | +----------+ | 0 | +----------+ 1 row in set (5.01 sec) MySQL [sakila]> MySQL [sakila]> SHOW TABLE tt DISTRIBUTIONS; +----------------+----------+------------+---------------------+-------------------+--------------------+-------------------+--------------------+--------------------------+-------------------------+--------------------------+------------------------+-----------------------+------------------------+ | PARTITION_NAME | STORE_ID | STORE_TYPE | REGION_LEADER_COUNT | REGION_PEER_COUNT | REGION_WRITE_BYTES | REGION_WRITE_KEYS | REGION_WRITE_QUERY | REGION_LEADER_READ_BYTES | REGION_LEADER_READ_KEYS | REGION_LEADER_READ_QUERY | REGION_PEER_READ_BYTES | REGION_PEER_READ_KEYS | REGION_PEER_READ_QUERY | +----------------+----------+------------+---------------------+-------------------+--------------------+-------------------+--------------------+--------------------------+-------------------------+--------------------------+------------------------+-----------------------+------------------------+ | p1 | 1 | tikv | 1 | 1 | 44195 | 108 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p1 | 2 | tikv | 0 | 1 | 44195 | 108 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p1 | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p2 | 1 | tikv | 1 | 1 | 44193 | 108 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p2 | 2 | tikv | 0 | 1 | 44193 | 108 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p2 | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p3 | 1 | tikv | 1 | 1 | 44193 | 108 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p3 | 2 | tikv | 0 | 1 | 44193 | 108 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p3 | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p4 | 2 | tikv | 0 | 1 | 44193 | 108 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p4 | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p4 | 1 | tikv | 1 | 1 | 44193 | 108 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p-max | 1 | tikv | 1 | 1 | 57695 | 616 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | | p-max | 2 | tikv | 0 | 1 | 57695 | 616 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | | p-max | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | +----------------+----------+------------+---------------------+-------------------+--------------------+-------------------+--------------------+--------------------------+-------------------------+--------------------------+------------------------+-----------------------+------------------------+ 15 rows in set (0.02 sec) MySQL [sakila]> SHOW TABLE tt REGIONS; +-----------+-----------+--------------------+-----------+-----------------+----------+------------+---------------+------------+----------------------+------------------+------------------------+------------------+ | REGION_ID | START_KEY | END_KEY | LEADER_ID | LEADER_STORE_ID | PEERS | SCATTERING | WRITTEN_BYTES | READ_BYTES | APPROXIMATE_SIZE(MB) | APPROXIMATE_KEYS | SCHEDULING_CONSTRAINTS | SCHEDULING_STATE | +-----------+-----------+--------------------+-----------+-----------------+----------+------------+---------------+------------+----------------------+------------------+------------------------+------------------+ | 270 | t_192_ | t_193_ | 272 | 1 | 271, 272 | 0 | 566907 | 0 | 1 | 0 | | | | 273 | t_193_ | t_194_ | 275 | 1 | 274, 275 | 0 | 566604 | 0 | 1 | 0 | | | | 276 | t_194_ | t_195_ | 278 | 1 | 277, 278 | 0 | 566616 | 0 | 1 | 0 | | | | 279 | t_195_ | t_196_ | 281 | 1 | 280, 281 | 0 | 566616 | 0 | 1 | 0 | | | | 28 | t_196_ | t_281474976710648_ | 31 | 1 | 29, 31 | 0 | 3461718 | 0 | 1 | 0 | | | +-----------+-----------+--------------------+-----------+-----------------+----------+------------+---------------+------------+----------------------+------------------+------------------------+------------------+ 5 rows in set (0.04 sec) MySQL [sakila]>对表的 Region 进行重新打散和调度，以实现表级别的均衡分布， 语法很简单，这里可以指定ENGINE 是tikv 或者tiflash。
MySQL [sakila]> DISTRIBUTE TABLE tt RULE = "leader-scatter" ENGINE = "tikv" TIMEOUT = "1h"; +--------+ | JOB_ID | +--------+ | 0 | +--------+ 1 row in set (0.06 sec) MySQL [sakila]> admin show ddl jobs; +--------+---------+----------------------------+--------------+--------------+-----------+----------+-----------+----------------------------+----------------------------+----------------------------+--------+----------+ | JOB_ID | DB_NAME | TABLE_NAME | JOB_TYPE | SCHEMA_STATE | SCHEMA_ID | TABLE_ID | ROW_COUNT | CREATE_TIME | START_TIME | END_TIME | STATE | COMMENTS | +--------+---------+----------------------------+--------------+--------------+-----------+----------+-----------+----------------------------+----------------------------+----------------------------+--------+----------+ | 197 | sakila | tt | create table | public | 145 | 191 | 0 | 2025-12-24 22:31:58.749000 | 2025-12-24 22:31:58.799000 | 2025-12-24 22:31:58.899000 | synced | | | 190 | sakila | t | create table | public | 145 | 187 | 0 | 2025-12-24 22:29:23.454000 | 2025-12-24 22:29:23.850000 | 2025-12-24 22:29:24.149000 | synced | | | 186 | sakila | t1 | create table | public | 145 | 185 | 0 | 2025-12-24 22:02:32.149000 | 2025-12-24 22:02:32.149000 | 2025-12-24 22:02:32.299000 | synced | | | 184 | sakila | actor_info | create view | public | 145 | 183 | 0 | 2025-12-24 21:55:04.699000 | 2025-12-24 21:55:04.699000 | 2025-12-24 21:55:04.749000 | synced | | | 182 | sakila | sales_by_film_category | create view | public | 145 | 181 | 0 | 2025-12-24 21:55:04.149000 | 2025-12-24 21:55:04.149000 | 2025-12-24 21:55:04.249000 | synced | | | 180 | sakila | nicer_but_slower_film_list | create view | public | 145 | 179 | 0 | 2025-12-24 21:55:03.600000 | 2025-12-24 21:55:03.649000 | 2025-12-24 21:55:03.699000 | synced | | | 178 | sakila | film_list | create view | public | 145 | 177 | 0 | 2025-12-24 21:55:03.049000 | 2025-12-24 21:55:03.101000 | 2025-12-24 21:55:03.149000 | synced | | | 176 | sakila | store | create table | public | 145 | 175 | 0 | 2025-12-24 21:55:02.549000 | 2025-12-24 21:55:02.549000 | 2025-12-24 21:55:02.699000 | synced | | | 174 | sakila | staff | create table | public | 145 | 173 | 0 | 2025-12-24 21:55:01.999000 | 2025-12-24 21:55:02.049000 | 2025-12-24 21:55:02.199000 | synced | | | 172 | sakila | rental | create table | public | 145 | 171 | 0 | 2025-12-24 21:55:01.451000 | 2025-12-24 21:55:01.499000 | 2025-12-24 21:55:01.649000 | synced | | +--------+---------+----------------------------+--------------+--------------+-----------+----------+-----------+----------------------------+----------------------------+----------------------------+--------+----------+ 10 rows in set (0.02 sec) MySQL [sakila]> SHOW TABLE tt DISTRIBUTIONS; +----------------+----------+------------+---------------------+-------------------+--------------------+-------------------+--------------------+--------------------------+-------------------------+--------------------------+------------------------+-----------------------+------------------------+ | PARTITION_NAME | STORE_ID | STORE_TYPE | REGION_LEADER_COUNT | REGION_PEER_COUNT | REGION_WRITE_BYTES | REGION_WRITE_KEYS | REGION_WRITE_QUERY | REGION_LEADER_READ_BYTES | REGION_LEADER_READ_KEYS | REGION_LEADER_READ_QUERY | REGION_PEER_READ_BYTES | REGION_PEER_READ_KEYS | REGION_PEER_READ_QUERY | +----------------+----------+------------+---------------------+-------------------+--------------------+-------------------+--------------------+--------------------------+-------------------------+--------------------------+------------------------+-----------------------+------------------------+ | p1 | 1 | tikv | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p1 | 2 | tikv | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p1 | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p2 | 1 | tikv | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p2 | 2 | tikv | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p2 | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p3 | 1 | tikv | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p3 | 2 | tikv | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p3 | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p4 | 1 | tikv | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p4 | 2 | tikv | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p4 | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p-max | 1 | tikv | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p-max | 2 | tikv | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | | p-max | 7 | tikv | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | +----------------+----------+------------+---------------------+-------------------+--------------------+-------------------+--------------------+--------------------------+-------------------------+--------------------------+------------------------+-----------------------+------------------------+ 15 rows in set (0.01 sec) MySQL [sakila]>注意事项
DISTRIBUTE TABLE语句在重新调度表中的 Region 时，可能会受到 PD 热点调度器的影响。调度完成后，随着时间推移，表的 Region 分布可能再次失衡。
