索引的选择

从存储层读取数据是 SQL 计算过程中最为耗时的部分之一，TiDB 目前支持从不同的存储和不同的索引中读取数据，索引选择得是否合理将很大程度上决定一个查询的运行速度。

本章节将介绍 TiDB 如何选择索引去读入数据，以及相关的一些控制索引选择的方式。

读表

在介绍索引的选择之前，首先要了解 TiDB 有哪些读表的方式，这些方式的触发条件是什么，不同方式有什么区别，各有什么优劣。

读表算子

读表算子 触发条件 适用场景 说明
PointGet/BatchPointGet 读表的范围是一个或多个单点范围 任何场景 如果能被触发，通常被认为是最快的算子，因为其直接调用 kvget 的接口进行计算，不走 coprocessor
TableReader 无 任何场景 该 TableReader 算子用于 TiKV。从 TiKV 端直接扫描表数据，一般被认为是效率最低的算子，除非在 _tidb_rowid 这一列上存在范围查询，或者无其他可以选择的读表算子时，才会选择这个算子
TableReader 表在 TiFlash 节点上存在副本 需要读取的列比较少，但是需要计算的行很多 该 TableReader 算子用于 TiFlash。TiFlash 是列式存储，如果需要对少量的列和大量的行进行计算，一般会选择这个算子
IndexReader 表有一个或多个索引，且计算所需的列被包含在索引里 存在较小的索引上的范围查询，或者对索引列有顺序需求的时候 当存在多个索引的时候，会根据估算代价选择合理的索引
IndexLookupReader 表有一个或多个索引，且计算所需的列不完全被包含在索引里 同 IndexReader 因为计算列不完全被包含在索引里，所以读完索引后需要回表，这里会比 IndexReader 多一些开销
IndexMerge 表有多个索引或多值索引 使用多值索引或同时使用多个索引的时候 可以通过 optimizer hints 指定使用该算子，或让优化器根据代价估算自动选择该算子，参见用 EXPLAIN 查看索引合并的 SQL 执行计划

索引的选择 - 图1

注意

TableReader 是基于 _tidb_rowid 的索引，TiFlash 是列存索引，所以索引的选择即是读表算子的选择。

索引的选择

TiDB 基于规则或基于代价来选择索引。基于的规则包括前置规则和 Skyline-Pruning。在选择索引时，TiDB 会先尝试前置规则。如果存在索引满足某一条前置规则，则直接选择该索引。否则，TiDB 会采用 Skyline-Pruning 来排除不合适的索引，然后基于每个读表算子的代价估算，选择代价最小的索引。

基于规则选择

前置规则

TiDB 采用如下的启发式前置规则来选择索引：

  • 规则 1：如果存在索引满足“唯一性索引全匹配 + 不需要回表（即该索引生成的计划是 IndexReader）”时，直接选择该索引。
  • 规则 2：如果存在索引满足“唯一性索引全匹配 + 需要回表（即该索引生成的计划是 IndexLookupReader）”时，选择满足该条件且回表行数最小的索引作为候选索引。
  • 规则 3：如果存在索引满足“普通索引不需要回表 + 读取行数小于一定阈值”时，选择满足该条件且读取行数最小的索引作为候选索引。
  • 规则 4：如果规则 2 和 3 之中仅选出一条候选索引，则选择该候选索引。如果规则 2 和 3 均选出候选索引，则选择读取行数（读索引行数 + 回表行数）较小的索引。

上述规则中的“索引全匹配”指每个索引列上均存在等值条件。在执行 EXPLAIN FORMAT = 'verbose' ... 语句时，如果前置规则匹配了某一索引，TiDB 会输出一条 NOTE 级别的 warning 提示该索引匹配了前置规则。

在以下示例中，因为索引 idx_b 满足规则 2 中“唯一性索引全匹配 + 需要回表”的条件，TiDB 选择索引 idx_b 作为访问路径，SHOW WARNING 返回了索引 idx_b 命中前置规则的提示。

1. mysql> CREATE TABLE t(a INT PRIMARY KEY, b INT, c INT, UNIQUE INDEX idx_b(b));
2. Query OK, 0 rows affected (0.01 sec)

4. mysql> EXPLAIN FORMAT = 'verbose' SELECT b, c FROM t WHERE b = 3 OR b = 6;
5. +-------------------+---------+---------+------+-------------------------+------------------------------+
6. | id                | estRows | estCost | task | access object           | operator info                |
7. +-------------------+---------+---------+------+-------------------------+------------------------------+
8. | Batch_Point_Get_5 | 2.00    | 8.80    | root | table:t, index:idx_b(b) | keep order:false, desc:false |
9. +-------------------+---------+---------+------+-------------------------+------------------------------+
10. 1 row in set, 1 warning (0.00 sec)

12. mysql> SHOW WARNINGS;
13. +-------+------+-------------------------------------------------------------------------------------------+
14. | Level | Code | Message                                                                                   |
15. +-------+------+-------------------------------------------------------------------------------------------+
16. | Note  | 1105 | unique index idx_b of t is selected since the path only has point ranges with double scan |
17. +-------+------+-------------------------------------------------------------------------------------------+
18. 1 row in set (0.00 sec)

Skyline-Pruning

Skyline-Pruning 是一个针对索引的启发式过滤规则，能降低错误估算导致选错索引的概率。Skyline-Pruning 从以下三个维度衡量一个索引的优劣：

  • 索引的列涵盖了多少访问条件。“访问条件”指的是可以转化为某列范围的 where 条件，如果某个索引的列集合涵盖的访问条件越多，那么它在这个维度上更优。
  • 选择该索引读表时，是否需要回表（即该索引生成的计划是 IndexReader 还是 IndexLookupReader）。不用回表的索引在这个维度上优于需要回表的索引。如果均需要回表，则比较索引的列涵盖了多少过滤条件。过滤条件指的是可以根据索引判断的 where 条件。如果某个索引的列集合涵盖的访问条件越多，则回表数量越少，那么它在这个维度上越优。
  • 选择该索引是否能满足一定的顺序。因为索引的读取可以保证某些列集合的顺序，所以满足查询要求顺序的索引在这个维度上优于不满足的索引。

对于以上三种维度，如果索引 idx_a 在这三个维度上都不比 idx_b 差，且有一个维度比 idx_b 好，那么 TiDB 会优先选择 idx_a。在执行 EXPLAIN FORMAT = 'verbose' ... 语句时，如果 Skyline-Pruning 排除了某些索引，TiDB 会输出一条 NOTE 级别的 warning 提示哪些索引在 Skyline-Pruning 排除之后保留下来。

在以下示例中，索引 idx_b 和 idx_e 均劣于 idx_b_c，因而被 Skyline-Pruning 排除，SHOW WARNING 的返回结果显示了经过 Skyline-Pruning 后剩余的索引。

1. mysql> CREATE TABLE t(a INT PRIMARY KEY, b INT, c INT, d INT, e INT, INDEX idx_b(b), INDEX idx_b_c(b, c), INDEX idx_e(e));
2. Query OK, 0 rows affected (0.01 sec)

4. mysql> EXPLAIN FORMAT = 'verbose' SELECT * FROM t WHERE b = 2 AND c > 4;
5. +-------------------------------+---------+---------+-----------+------------------------------+----------------------------------------------------+
6. | id                            | estRows | estCost | task      | access object                | operator info                                      |
7. +-------------------------------+---------+---------+-----------+------------------------------+----------------------------------------------------+
8. | IndexLookUp_10                | 33.33   | 738.29  | root      |                              |                                                    |
9. | ├─IndexRangeScan_8(Build)     | 33.33   | 2370.00 | cop[tikv] | table:t, index:idx_b_c(b, c) | range:(2 4,2 +inf], keep order:false, stats:pseudo |
10. | └─TableRowIDScan_9(Probe)     | 33.33   | 2370.00 | cop[tikv] | table:t                      | keep order:false, stats:pseudo                     |
11. +-------------------------------+---------+---------+-----------+------------------------------+----------------------------------------------------+
12. 3 rows in set, 1 warning (0.00 sec)

14. mysql> SHOW WARNINGS;
15. +-------+------+------------------------------------------------------------------------------------------+
16. | Level | Code | Message                                                                                  |
17. +-------+------+------------------------------------------------------------------------------------------+
18. | Note  | 1105 | [t,idx_b_c] remain after pruning paths for t given Prop{SortItems: [], TaskTp: rootTask} |
19. +-------+------+------------------------------------------------------------------------------------------+
20. 1 row in set (0.00 sec)

基于代价选择

在使用 Skyline-Pruning 规则排除了不合适的索引之后，索引的选择完全基于代价估算，读表的代价估算需要考虑以下几个方面：

  • 索引的每行数据在存储层的平均长度。
  • 索引生成的查询范围的行数量。
  • 索引的回表代价。
  • 索引查询时的范围数量。

根据这些因子和代价模型，优化器会选择一个代价最低的索引进行读表。

代价选择调优的常见问题

  1. 估算的行数量不准确？

一般是统计信息过期或者准确度不够造成的，可以重新执行 ANALYZE TABLE 或者修改 ANALYZE TABLE 的参数。 2. 统计信息准确，为什么读 TiFlash 更快，而优化器选择了 TiKV？

目前区别 TiFlash 和 TiKV 的代价模型还比较粗糙，可以调小 tidb_opt_seek_factor 的值，让优化器倾向于选择 TiFlash。 3. 统计信息准确，某个索引要回表，但是它比另一个不用回表的索引实际执行更快，为什么选择了不用回表的索引？

碰到这种情况，可能是代价估算时对于回表的代价计算得过大，可以调小 tidb_opt_network_factor，降低回表的代价。

控制索引的选择

通过 Optimizer Hints 可以实现单条查询对索引选择的控制。

  • USE_INDEX/IGNORE_INDEX 可以强制优化器使用/不使用某些索引。FORCE_INDEX 和 USE_INDEX 的作用相同。
  • READ_FROM_STORAGE 可以强制优化器对于某些表选择 TiKV/TiFlash 的存储引擎进行查询。

使用多值索引

多值索引和普通索引有所不同，TiDB 目前只会使用 IndexMerge 来访问多值索引。因此要想使用多值索引进行数据访问，请确保 tidb_enable_index_merge 被设置为 ON。

多值索引的使用限制请参考 CREATE INDEX。

支持多值索引的场景

目前 TiDB 支持将 json_member_of、json_contains 和 json_overlaps 条件自动转换成 IndexMerge 来访问多值索引。既可以依赖优化器根据代价自动选择，也可通过 use_index_merge optimizer hint 或 use_index 指定选择多值索引，见下面例子：

1. mysql> CREATE TABLE t1 (j JSON, INDEX idx((CAST(j->'$.path' AS SIGNED ARRAY)))); -- 使用 '$.path' 作为路径创建多值索引
2. Query OK, 0 rows affected (0.04 sec)

4. mysql> EXPLAIN SELECT /*+ use_index_merge(t1, idx) */ * FROM t1 WHERE (1 MEMBER OF (j->'$.path'));
5. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+------------------------------------------------------------------------+
6. | id                              | estRows | task      | access object                                                               | operator info                                                          |
7. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+------------------------------------------------------------------------+
8. | Selection_5                     | 8000.00 | root      |                                                                             | json_memberof(cast(1, json BINARY), json_extract(test.t1.j, "$.path")) |
9. | └─IndexMerge_8                  | 10.00   | root      |                                                                             | type: union                                                            |
10. |   ├─IndexRangeScan_6(Build)     | 10.00   | cop[tikv] | table:t1, index:idx(cast(json_extract(`j`, _utf8'$.path') as signed array)) | range:[1,1], keep order:false, stats:pseudo                            |
11. |   └─TableRowIDScan_7(Probe)     | 10.00   | cop[tikv] | table:t1                                                                    | keep order:false, stats:pseudo                                         |
12. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+------------------------------------------------------------------------+
13. 4 rows in set, 1 warning (0.00 sec)

15. mysql> EXPLAIN SELECT /*+ use_index_merge(t1, idx) */ * FROM t1 WHERE JSON_CONTAINS((j->'$.path'), '[1, 2, 3]');
16. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+
17. | id                            | estRows | task      | access object                                                               | operator info                               |
18. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+
19. | IndexMerge_9                  | 10.00   | root      |                                                                             | type: intersection                          |
20. | ├─IndexRangeScan_5(Build)     | 10.00   | cop[tikv] | table:t1, index:idx(cast(json_extract(`j`, _utf8'$.path') as signed array)) | range:[1,1], keep order:false, stats:pseudo |
21. | ├─IndexRangeScan_6(Build)     | 10.00   | cop[tikv] | table:t1, index:idx(cast(json_extract(`j`, _utf8'$.path') as signed array)) | range:[2,2], keep order:false, stats:pseudo |
22. | ├─IndexRangeScan_7(Build)     | 10.00   | cop[tikv] | table:t1, index:idx(cast(json_extract(`j`, _utf8'$.path') as signed array)) | range:[3,3], keep order:false, stats:pseudo |
23. | └─TableRowIDScan_8(Probe)     | 10.00   | cop[tikv] | table:t1                                                                    | keep order:false, stats:pseudo              |
24. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+
25. 5 rows in set (0.00 sec)

27. mysql> EXPLAIN SELECT /*+ use_index_merge(t1, idx) */ * FROM t1 WHERE JSON_OVERLAPS((j->'$.path'), '[1, 2, 3]');
28. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+----------------------------------------------------------------------------------+
29. | id                              | estRows | task      | access object                                                               | operator info                                                                    |
30. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+----------------------------------------------------------------------------------+
31. | Selection_5                     | 8000.00 | root      |                                                                             | json_overlaps(json_extract(test.t1.j, "$.path"), cast("[1, 2, 3]", json BINARY)) |
32. | └─IndexMerge_10                 | 10.00   | root      |                                                                             | type: union                                                                      |
33. |   ├─IndexRangeScan_6(Build)     | 10.00   | cop[tikv] | table:t1, index:idx(cast(json_extract(`j`, _utf8'$.path') as signed array)) | range:[1,1], keep order:false, stats:pseudo                                      |
34. |   ├─IndexRangeScan_7(Build)     | 10.00   | cop[tikv] | table:t1, index:idx(cast(json_extract(`j`, _utf8'$.path') as signed array)) | range:[2,2], keep order:false, stats:pseudo                                      |
35. |   ├─IndexRangeScan_8(Build)     | 10.00   | cop[tikv] | table:t1, index:idx(cast(json_extract(`j`, _utf8'$.path') as signed array)) | range:[3,3], keep order:false, stats:pseudo                                      |
36. |   └─TableRowIDScan_9(Probe)     | 10.00   | cop[tikv] | table:t1                                                                    | keep order:false, stats:pseudo                                                   |
37. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+----------------------------------------------------------------------------------+
38. 6 rows in set, 1 warning (0.00 sec)

复合的多值索引，也一样可以使用 IndexMerge 进行访问：

1. CREATE TABLE t2 (a INT, j JSON, b INT, k JSON, INDEX idx(a, (CAST(j->'$.path' AS SIGNED ARRAY)), b), INDEX idx2(b, (CAST(k->'$.path' AS SIGNED ARRAY))));
2. EXPLAIN SELECT /*+ use_index_merge(t2, idx) */ * FROM t2 WHERE a=1 AND (1 MEMBER OF (j->'$.path')) AND b=2;
3. EXPLAIN SELECT /*+ use_index_merge(t2, idx) */ * FROM t2 WHERE a=1 AND JSON_CONTAINS((j->'$.path'), '[1, 2, 3]');
4. EXPLAIN SELECT /*+ use_index_merge(t2, idx) */ * FROM t2 WHERE a=1 AND JSON_OVERLAPS((j->'$.path'), '[1, 2, 3]');
5. EXPLAIN SELECT /*+ use_index_merge(t2, idx, idx2) */ * FROM t2 WHERE (a=1 AND 1 member of (j->'$.path')) AND (b=1 AND 2 member of (k->'$.path'));

1. > EXPLAIN SELECT /*+ use_index_merge(t2, idx) */ * FROM t2 WHERE a=1 AND (1 MEMBER OF (j->'$.path')) AND b=2;
2. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-----------------------------------------------------+
3. | id                            | estRows | task      | access object                                                                     | operator info                                       |
4. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-----------------------------------------------------+
5. | IndexMerge_7                  | 0.00    | root      |                                                                                   | type: union                                         |
6. | ├─IndexRangeScan_5(Build)     | 0.00    | cop[tikv] | table:t2, index:idx(a, cast(json_extract(`j`, _utf8'$.path') as signed array), b) | range:[1 1 2,1 1 2], keep order:false, stats:pseudo |
7. | └─TableRowIDScan_6(Probe)     | 0.00    | cop[tikv] | table:t2                                                                          | keep order:false, stats:pseudo                      |
8. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-----------------------------------------------------+

10. > EXPLAIN SELECT /*+ use_index_merge(t2, idx) */ * FROM t2 WHERE a=1 AND JSON_CONTAINS((j->'$.path'), '[1, 2, 3]');
11. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-------------------------------------------------+
12. | id                            | estRows | task      | access object                                                                     | operator info                                   |
13. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-------------------------------------------------+
14. | IndexMerge_9                  | 0.00    | root      |                                                                                   | type: intersection                              |
15. | ├─IndexRangeScan_5(Build)     | 0.10    | cop[tikv] | table:t2, index:idx(a, cast(json_extract(`j`, _utf8'$.path') as signed array), b) | range:[1 1,1 1], keep order:false, stats:pseudo |
16. | ├─IndexRangeScan_6(Build)     | 0.10    | cop[tikv] | table:t2, index:idx(a, cast(json_extract(`j`, _utf8'$.path') as signed array), b) | range:[1 2,1 2], keep order:false, stats:pseudo |
17. | ├─IndexRangeScan_7(Build)     | 0.10    | cop[tikv] | table:t2, index:idx(a, cast(json_extract(`j`, _utf8'$.path') as signed array), b) | range:[1 3,1 3], keep order:false, stats:pseudo |
18. | └─TableRowIDScan_8(Probe)     | 0.00    | cop[tikv] | table:t2                                                                          | keep order:false, stats:pseudo                  |
19. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-------------------------------------------------+

21. > EXPLAIN SELECT /*+ use_index_merge(t2, idx) */ * FROM t2 WHERE a=1 AND JSON_OVERLAPS((j->'$.path'), '[1, 2, 3]');
22. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------------+----------------------------------------------------------------------------------+
23. | id                              | estRows | task      | access object                                                                     | operator info                                                                    |
24. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------------+----------------------------------------------------------------------------------+
25. | Selection_5                     | 0.24    | root      |                                                                                   | json_overlaps(json_extract(test.t2.j, "$.path"), cast("[1, 2, 3]", json BINARY)) |
26. | └─IndexMerge_10                 | 0.30    | root      |                                                                                   | type: union                                                                      |
27. |   ├─IndexRangeScan_6(Build)     | 0.10    | cop[tikv] | table:t2, index:idx(a, cast(json_extract(`j`, _utf8'$.path') as signed array), b) | range:[1 1,1 1], keep order:false, stats:pseudo                                  |
28. |   ├─IndexRangeScan_7(Build)     | 0.10    | cop[tikv] | table:t2, index:idx(a, cast(json_extract(`j`, _utf8'$.path') as signed array), b) | range:[1 2,1 2], keep order:false, stats:pseudo                                  |
29. |   ├─IndexRangeScan_8(Build)     | 0.10    | cop[tikv] | table:t2, index:idx(a, cast(json_extract(`j`, _utf8'$.path') as signed array), b) | range:[1 3,1 3], keep order:false, stats:pseudo                                  |
30. |   └─TableRowIDScan_9(Probe)     | 0.30    | cop[tikv] | table:t2                                                                          | keep order:false, stats:pseudo                                                   |
31. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------------+----------------------------------------------------------------------------------+

33. > EXPLAIN SELECT /*+ use_index_merge(t2, idx, idx2) */ * FROM t2 WHERE (a=1 AND 1 member of (j->'$.path')) AND (b=1 AND 2 member of (k->'$.path'));
34. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-----------------------------------------------------+
35. | id                            | estRows | task      | access object                                                                     | operator info                                       |
36. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-----------------------------------------------------+
37. | IndexMerge_8                  | 0.00    | root      |                                                                                   | type: intersection                                  |
38. | ├─IndexRangeScan_5(Build)     | 0.00    | cop[tikv] | table:t2, index:idx(a, cast(json_extract(`j`, _utf8'$.path') as signed array), b) | range:[1 1 1,1 1 1], keep order:false, stats:pseudo |
39. | ├─IndexRangeScan_6(Build)     | 0.10    | cop[tikv] | table:t2, index:idx2(b, cast(json_extract(`k`, _utf8'$.path') as signed array))   | range:[1 2,1 2], keep order:false, stats:pseudo     |
40. | └─TableRowIDScan_7(Probe)     | 0.00    | cop[tikv] | table:t2                                                                          | keep order:false, stats:pseudo                      |
41. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------------+-----------------------------------------------------+

同时，多值索引可以和普通索引共同组成 IndexMerge 使用。例如：

1. CREATE TABLE t3(j1 JSON, j2 JSON, a INT, INDEX k1((CAST(j1->'$.path' AS SIGNED ARRAY))), INDEX k2((CAST(j2->'$.path' AS SIGNED ARRAY))), INDEX ka(a));
2. EXPLAIN SELECT /*+ use_index_merge(t3, k1, k2, ka) */ * FROM t3 WHERE 1 member of (j1->'$.path') OR a = 3;
3. EXPLAIN SELECT /*+ use_index_merge(t3, k1, k2, ka) */ * FROM t3 WHERE 1 member of (j1->'$.path') AND 2 member of (j2->'$.path') AND (a = 3);

1. > EXPLAIN SELECT /*+ use_index_merge(t3, k1, k2, ka) */ * FROM t3 WHERE 1 member of (j1->'$.path') OR a = 3;
2. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+
3. | id                            | estRows | task      | access object                                                               | operator info                               |
4. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+
5. | IndexMerge_8                  | 19.99   | root      |                                                                             | type: union                                 |
6. | ├─IndexRangeScan_5(Build)     | 10.00   | cop[tikv] | table:t3, index:k1(cast(json_extract(`j1`, _utf8'$.path') as signed array)) | range:[1,1], keep order:false, stats:pseudo |
7. | ├─IndexRangeScan_6(Build)     | 10.00   | cop[tikv] | table:t3, index:ka(a)                                                       | range:[3,3], keep order:false, stats:pseudo |
8. | └─TableRowIDScan_7(Probe)     | 19.99   | cop[tikv] | table:t3                                                                    | keep order:false, stats:pseudo              |
9. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+

11. > EXPLAIN SELECT /*+ use_index_merge(t3, k1, k2, ka) */ * FROM t3 WHERE 1 member of (j1->'$.path') AND 2 member of (j2->'$.path') AND (a = 3);
12. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+
13. | id                            | estRows | task      | access object                                                               | operator info                               |
14. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+
15. | IndexMerge_9                  | 0.00    | root      |                                                                             | type: intersection                          |
16. | ├─IndexRangeScan_5(Build)     | 10.00   | cop[tikv] | table:t3, index:ka(a)                                                       | range:[3,3], keep order:false, stats:pseudo |
17. | ├─IndexRangeScan_6(Build)     | 10.00   | cop[tikv] | table:t3, index:k1(cast(json_extract(`j1`, _utf8'$.path') as signed array)) | range:[1,1], keep order:false, stats:pseudo |
18. | ├─IndexRangeScan_7(Build)     | 10.00   | cop[tikv] | table:t3, index:k2(cast(json_extract(`j2`, _utf8'$.path') as signed array)) | range:[2,2], keep order:false, stats:pseudo |
19. | └─TableRowIDScan_8(Probe)     | 0.00    | cop[tikv] | table:t3                                                                    | keep order:false, stats:pseudo              |
20. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+

对于由 OR 或 AND 连接而成的多个 json_member_of、json_contains 或 json_overlaps 条件，需要满足以下条件才能使用 IndexMerge 访问：

1. CREATE TABLE t4(a INT, j JSON, INDEX mvi1((CAST(j->'$.a' AS UNSIGNED ARRAY))), INDEX mvi2((CAST(j->'$.b' AS UNSIGNED ARRAY))));
  • 对于由 OR 连接而成的条件，其中每个子条件都需要分别能够使用 IndexMerge 访问。例如：

1. EXPLAIN SELECT /*+ use_index_merge(t4, mvi1) */ * FROM t4 WHERE json_overlaps(j->'$.a', '[1, 2]') OR json_overlaps(j->'$.a', '[3, 4]'); 2. EXPLAIN SELECT /*+ use_index_merge(t4, mvi1) */ * FROM t4 WHERE json_overlaps(j->'$.a', '[1, 2]') OR json_length(j->'$.a') = 3; 3. SHOW WARNINGS;

`` 1. > EXPLAIN SELECT /*+ use_index_merge(t4, mvi1) */ * FROM t4 WHERE json_overlaps(j->'$.a', '[1, 2]') OR json_overlaps(j->'$.a', '[3, 4]'); 2. +----------------------------------+---------+-----------+-----------------------------------------------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------+ 3. | id | estRows | task | access object | operator info | 4. +----------------------------------+---------+-----------+-----------------------------------------------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------+ 5. | Selection_5 | 31.95 | root | | or(json_overlaps(json_extract(test.t4.j, "$.a"), cast("[1, 2]", json BINARY)), json_overlaps(json_extract(test.t4.j, "$.a"), cast("[3, 4]", json BINARY))) | 6. | └─IndexMerge_11 | 39.94 | root | | type: union | 7. | ├─IndexRangeScan_6(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[1,1], keep order:false, stats:pseudo | 8. | ├─IndexRangeScan_7(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[2,2], keep order:false, stats:pseudo | 9. | ├─IndexRangeScan_8(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[3,3], keep order:false, stats:pseudo | 10. | ├─IndexRangeScan_9(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j`, _utf8'$.a') as unsigned array)) | range:[4,4], keep order:false, stats:pseudo | 11. | └─TableRowIDScan_10(Probe) | 39.94 | cop[tikv] | table:t4 | keep order:false, stats:pseudo | 12. +----------------------------------+---------+-----------+-----------------------------------------------------------------------------+------------------------------------------------------------------------------------------------------------------------------------------------------------+

  1. -- json_length(j->'$.a') = 3 无法使用 IndexMerge 访问，因此该 SQL 整体无法使用 IndexMerge 访问

  2. EXPLAIN SELECT /+ use_index_merge(t4, mvi1) / * FROM t4 WHERE json_overlaps(j->'$.a', '[1, 2]') OR json_length(j->'$.a') = 3;

  3. +-------------------------+----------+-----------+---------------+------------------------------------------------------------------------------------------------------------------------------------+
  4. | id | estRows | task | access object | operator info |
  5. +-------------------------+----------+-----------+---------------+------------------------------------------------------------------------------------------------------------------------------------+
  6. | Selection_5 | 8000.00 | root | | or(json_overlaps(json_extract(test.t4.j, "$.a"), cast("[1, 2]", json BINARY)), eq(json_length(json_extract(test.t4.j, "$.a")), 3)) |
  7. | └─TableReader_7 | 10000.00 | root | | data:TableFullScan_6 |
  8. | └─TableFullScan_6 | 10000.00 | cop[tikv] | table:t4 | keep order:false, stats:pseudo |

  9. +-------------------------+----------+-----------+---------------+------------------------------------------------------------------------------------------------------------------------------------+

  10. SHOW WARNINGS;

  11. +---------+------+----------------------------+
  12. | Level | Code | Message |
  13. +---------+------+----------------------------+
  14. | Warning | 1105 | IndexMerge is inapplicable |
  15. +---------+------+----------------------------+ ```
  16. 对于由 AND 连接而成的条件，其中一部分子条件需要分别能够使用 IndexMerge 访问。最终使用 IndexMerge 访问时，也只能利用这一部分条件。例如：

1. EXPLAIN SELECT /*+ use_index_merge(t4, mvi1) */ * FROM t4 WHERE json_contains(j->'$.a', '[1, 2]') AND json_contains(j->'$.a', '[3, 4]'); 2. EXPLAIN SELECT /*+ use_index_merge(t4, mvi1) */ * FROM t4 WHERE json_contains(j->'$.a', '[1, 2]') AND json_contains(j->'$.a', '[3, 4]') AND json_length(j->'$.a') = 2;

`` 1. > EXPLAIN SELECT /*+ use_index_merge(t4, mvi1) */ * FROM t4 WHERE json_contains(j->'$.a', '[1, 2]') AND json_contains(j->'$.a', '[3, 4]'); 2. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+ 3. | id | estRows | task | access object | operator info | 4. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+ 5. | IndexMerge_10 | 0.00 | root | | type: intersection | 6. | ├─IndexRangeScan_5(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[1,1], keep order:false, stats:pseudo | 7. | ├─IndexRangeScan_6(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[2,2], keep order:false, stats:pseudo | 8. | ├─IndexRangeScan_7(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[3,3], keep order:false, stats:pseudo | 9. | ├─IndexRangeScan_8(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j`, _utf8'$.a') as unsigned array)) | range:[4,4], keep order:false, stats:pseudo | 10. | └─TableRowIDScan_9(Probe) | 0.00 | cop[tikv] | table:t4 | keep order:false, stats:pseudo | 11. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+

  1. -- json_length(j->'$.a') = 3 无法使用 IndexMerge 访问，因此只有另外两个 json_contains 表达式可以使用 IndexMerge 访问，而 json_length(j->'$.a') = 3 成为一个单独的 Selection 算子

  2. EXPLAIN SELECT /+ use_index_merge(t4, mvi1) / * FROM t4 WHERE json_contains(j->'$.a', '[1, 2]') AND json_contains(j->'$.a', '[3, 4]') AND json_length(j->'$.a') = 2;

  3. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+----------------------------------------------------+
  4. | id | estRows | task | access object | operator info |
  5. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+----------------------------------------------------+
  6. | IndexMerge_11 | 0.00 | root | | type: intersection |
  7. | ├─IndexRangeScan_5(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[1,1], keep order:false, stats:pseudo |
  8. | ├─IndexRangeScan_6(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[2,2], keep order:false, stats:pseudo |
  9. | ├─IndexRangeScan_7(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[3,3], keep order:false, stats:pseudo |
  10. | ├─IndexRangeScan_8(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[4,4], keep order:false, stats:pseudo |
  11. | └─Selection_10(Probe) | 0.00 | cop[tikv] | | eq(json_length(json_extract(test.t4.j, "$.a")), 2) |
  12. | └─TableRowIDScan_9 | 0.00 | cop[tikv] | table:t4 | keep order:false, stats:pseudo |
  13. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+----------------------------------------------------+ ```

  14. 用于构成整体 IndexMerge 计划的每个子条件，需要分别符合用于连接的 OR 或 AND 的语义。具体如下：

  15. json_contains 由 AND 连接，符合语义。例如：

    1. EXPLAIN SELECT /*+ use_index_merge(t4, mvi1, mvi2) */ * FROM t4 WHERE json_contains(j->'$.a', '[1]') AND json_contains(j->'$.b', '[2, 3]'); 2. EXPLAIN SELECT /*+ use_index_merge(t4, mvi1, mvi2) */ * FROM t4 WHERE json_contains(j->'$.a', '[1]') OR json_contains(j->'$.b', '[2, 3]');

    `` 1. > EXPLAIN SELECT /*+ use_index_merge(t4, mvi1, mvi2) */ * FROM t4 WHERE json_contains(j->'$.a', '[1]') AND json_contains(j->'$.b', '[2, 3]'); 2. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+ 3. | id | estRows | task | access object | operator info | 4. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+ 5. | IndexMerge_9 | 0.00 | root | | type: intersection | 6. | ├─IndexRangeScan_5(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[1,1], keep order:false, stats:pseudo | 7. | ├─IndexRangeScan_6(Build) | 10.00 | cop[tikv] | table:t4, index:mvi2(cast(json_extract(j, _utf8'$.b') as unsigned array)) | range:[2,2], keep order:false, stats:pseudo | 8. | ├─IndexRangeScan_7(Build) | 10.00 | cop[tikv] | table:t4, index:mvi2(cast(json_extract(j`, _utf8'$.b') as unsigned array)) | range:[3,3], keep order:false, stats:pseudo | 9. | └─TableRowIDScan_8(Probe) | 0.00 | cop[tikv] | table:t4 | keep order:false, stats:pseudo | 10. +-------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------+

    1. -- 不符合语义，如前文所述，该 SQL 整体无法使用 IndexMerge 访问

    2. EXPLAIN SELECT /+ use_index_merge(t4, mvi1, mvi2) / * FROM t4 WHERE json_contains(j->'$.a', '[1]') OR json_contains(j->'$.b', '[2, 3]');

    3. +-------------------------+----------+-----------+---------------+---------------------------------------------------------------------------------------------------------------------------------------------------------+
    4. | id | estRows | task | access object | operator info |
    5. +-------------------------+----------+-----------+---------------+---------------------------------------------------------------------------------------------------------------------------------------------------------+
    6. | TableReader_7 | 10.01 | root | | data:Selection_6 |
    7. | └─Selection_6 | 10.01 | cop[tikv] | | or(json_contains(json_extract(test.t4.j, "$.a"), cast("[1]", json BINARY)), json_contains(json_extract(test.t4.j, "$.b"), cast("[2, 3]", json BINARY))) |
    8. | └─TableFullScan_5 | 10000.00 | cop[tikv] | table:t4 | keep order:false, stats:pseudo |
    9. +-------------------------+----------+-----------+---------------+---------------------------------------------------------------------------------------------------------------------------------------------------------+ ```
    10. json_overlaps 由 OR 连接，符合语义。例如：

    1. EXPLAIN SELECT /*+ use_index_merge(t4, mvi1, mvi2) */ * FROM t4 WHERE json_overlaps(j->'$.a', '[1]') OR json_overlaps(j->'$.b', '[2, 3]'); 2. EXPLAIN SELECT /*+ use_index_merge(t4, mvi1, mvi2) */ * FROM t4 WHERE json_overlaps(j->'$.a', '[1]') AND json_overlaps(j->'$.b', '[2, 3]');

    `` 1. > EXPLAIN SELECT /*+ use_index_merge(t4, mvi1, mvi2) */ * FROM t4 WHERE json_overlaps(j->'$.a', '[1]') OR json_overlaps(j->'$.b', '[2, 3]'); 2. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------+ 3. | id | estRows | task | access object | operator info | 4. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------+ 5. | Selection_5 | 23.98 | root | | or(json_overlaps(json_extract(test.t4.j, "$.a"), cast("[1]", json BINARY)), json_overlaps(json_extract(test.t4.j, "$.b"), cast("[2, 3]", json BINARY))) | 6. | └─IndexMerge_10 | 29.97 | root | | type: union | 7. | ├─IndexRangeScan_6(Build) | 10.00 | cop[tikv] | table:t4, index:mvi1(cast(json_extract(j, _utf8'$.a') as unsigned array)) | range:[1,1], keep order:false, stats:pseudo | 8. | ├─IndexRangeScan_7(Build) | 10.00 | cop[tikv] | table:t4, index:mvi2(cast(json_extract(j, _utf8'$.b') as unsigned array)) | range:[2,2], keep order:false, stats:pseudo | 9. | ├─IndexRangeScan_8(Build) | 10.00 | cop[tikv] | table:t4, index:mvi2(cast(json_extract(j`, _utf8'$.b') as unsigned array)) | range:[3,3], keep order:false, stats:pseudo | 10. | └─TableRowIDScan_9(Probe) | 29.97 | cop[tikv] | table:t4 | keep order:false, stats:pseudo | 11. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------------------------+

    1. -- 不符合语义，如前文所述，只能利用一部分条件使用 IndexMerge 访问

    2. EXPLAIN SELECT /+ use_index_merge(t4, mvi1, mvi2) / * FROM t4 WHERE json_overlaps(j->'$.a', '[1]') AND json_overlaps(j->'$.b', '[2, 3]');

    3. +---------------------------------+---------+-----------+-----------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------+
    4. | id

（本文超出站点单文档体积上限，已截断）