planner: add document about some new introduced hints (#14392) (#14857)

Author: ti-chi-bot

optimizer-hints.md

@@ -96,6 +96,14 @@ select /*+ MERGE_JOIN(t1, t2) */ * from t1, t2 where t1.id = t2.id;
 >
 > `TIDB_SMJ` is the alias for `MERGE_JOIN` in TiDB 3.0.x and earlier versions. If you are using any of these versions, you must apply the `TIDB_SMJ(t1_name [, tl_name ...])` syntax for the hint. For the later versions of TiDB, `TIDB_SMJ` and `MERGE_JOIN` are both valid names for the hint, but `MERGE_JOIN` is recommended.
 
+### NO_MERGE_JOIN(t1_name [, tl_name ...])
+
+The `NO_MERGE_JOIN(t1_name [, tl_name ...])` hint tells the optimizer not to use the sort-merge join algorithm for the given table(s). For example:
+
+```sql
+SELECT /*+ NO_MERGE_JOIN(t1, t2) */ * FROM t1, t2 WHERE t1.id = t2.id;
+```
+
 ### INL_JOIN(t1_name [, tl_name ...])
 
 The `INL_JOIN(t1_name [, tl_name ...])` hint tells the optimizer to use the index nested loop join algorithm for the given table(s). This algorithm might consume less system resources and take shorter processing time in some scenarios and might produce an opposite result in other scenarios. If the result set is less than 10,000 rows after the outer table is filtered by the `WHERE` condition, it is recommended to use this hint. For example:
@@ -112,10 +120,30 @@ The parameter(s) given in `INL_JOIN()` is the candidate table for the inner tabl
 >
 > `TIDB_INLJ` is the alias for `INL_JOIN` in TiDB 3.0.x and earlier versions. If you are using any of these versions, you must apply the `TIDB_INLJ(t1_name [, tl_name ...])` syntax for the hint. For the later versions of TiDB, `TIDB_INLJ` and `INL_JOIN` are both valid names for the hint, but `INL_JOIN` is recommended.
 
+### NO_INDEX_JOIN(t1_name [, tl_name ...])
+
+The `NO_INDEX_JOIN(t1_name [, tl_name ...])` hint tells the optimizer not to use the index nested loop join algorithm for the given table(s). For example:
+
+```sql
+SELECT /*+ NO_INDEX_JOIN(t1, t2) */ * FROM t1, t2 WHERE t1.id = t2.id;
+```
+
 ### INL_HASH_JOIN
 
 The `INL_HASH_JOIN(t1_name [, tl_name])` hint tells the optimizer to use the index nested loop hash join algorithm. The conditions for using this algorithm are the same with the conditions for using the index nested loop join algorithm. The difference between the two algorithms is that `INL_JOIN` creates a hash table on the joined inner table, but `INL_HASH_JOIN` creates a hash table on the joined outer table. `INL_HASH_JOIN` has a fixed limit on memory usage, while the memory used by `INL_JOIN` depends on the number of rows matched in the inner table.
 
+### NO_INDEX_HASH_JOIN(t1_name [, tl_name ...])
+
+The `NO_INDEX_HASH_JOIN(t1_name [, tl_name ...])` hint tells the optimizer not to use the index nested loop hash join algorithm for the given table(s).
+
+### INL_MERGE_JOIN
+
+The `INL_MERGE_JOIN(t1_name [, tl_name])` hint tells the optimizer to use the index nested loop merge join algorithm. The conditions for using this algorithm are the same with the conditions for using the index nested loop join algorithm.
+
+### NO_INDEX_MERGE_JOIN(t1_name [, tl_name ...])
+
+The `NO_INDEX_MERGE_JOIN(t1_name [, tl_name ...])` hint tells the optimizer not to use the index nested loop merge join algorithm for the given table(s).
+
 ### HASH_JOIN(t1_name [, tl_name ...])
 
 The `HASH_JOIN(t1_name [, tl_name ...])` hint tells the optimizer to use the hash join algorithm for the given table(s). This algorithm allows the query to be executed concurrently with multiple threads, which achieves a higher processing speed but consumes more memory. For example:
@@ -130,6 +158,14 @@ select /*+ HASH_JOIN(t1, t2) */ * from t1, t2 where t1.id = t2.id;
 >
 > `TIDB_HJ` is the alias for `HASH_JOIN` in TiDB 3.0.x and earlier versions. If you are using any of these versions, you must apply the `TIDB_HJ(t1_name [, tl_name ...])` syntax for the hint. For the later versions of TiDB, `TIDB_HJ` and `HASH_JOIN` are both valid names for the hint, but `HASH_JOIN` is recommended.
 
+### NO_HASH_JOIN(t1_name [, tl_name ...])
+
+The `NO_HASH_JOIN(t1_name [, tl_name ...])` hint tells the optimizer not to use the hash join algorithm for the given table(s). For example:
+
+```sql
+SELECT /*+ NO_HASH_JOIN(t1, t2) */ * FROM t1, t2 WHERE t1.id = t2.id;
+```
+
 ### HASH_JOIN_BUILD(t1_name [, tl_name ...])
 
 The `HASH_JOIN_BUILD(t1_name [, tl_name ...])` hint tells the optimizer to use the hash join algorithm on specified tables with these tables working as the build side. In this way, you can build hash tables using specific tables. For example: