数据库内核月报－ 2019/05 - MySQL · 引擎特性 · Skip Scan Range - 《数据库内核月报》

参考

MySQL从8.0.13版本开始支持一种新的range scan方式，称为Loose Skip Scan。该特性由Facebook贡献。我们知道在之前的版本中，如果要使用到索引进行扫描，条件必须满足索引前缀列，比如索引idx(col1,col2), 如果where条件只包含col2的话，是无法有效的使用idx的, 它需要扫描索引上所有的行，然后再根据col2上的条件过滤。

新的优化可以避免全量索引扫描，而是根据每个col1上的值+col2上的条件，启动多次range scan。每次range scan根据构建的key值直接在索引上定位，直接忽略了那些不满足条件的记录。

示例下例是从官方文档上摘取的例子:


1. root@test 11:03:28>CREATE TABLE t1 (f1 INT NOT NULL, f2 INT NOT NULL, PRIMARY KEY(f1, f2));
2. Query OK, 0 rows affected (0.00 sec)

4. root@test 11:03:29>INSERT INTO t1 VALUES
5. ->   (1,1), (1,2), (1,3), (1,4), (1,5),
6. ->   (2,1), (2,2), (2,3), (2,4), (2,5);
7. Query OK, 10 rows affected (0.00 sec)
8. Records: 10  Duplicates: 0  Warnings: 0

10. root@test 11:03:29>INSERT INTO t1 SELECT f1, f2 + 5 FROM t1;
11. Query OK, 10 rows affected (0.00 sec)
12. Records: 10  Duplicates: 0  Warnings: 0

14. root@test 11:03:29>INSERT INTO t1 SELECT f1, f2 + 10 FROM t1;
15. Query OK, 20 rows affected (0.00 sec)
16. Records: 20  Duplicates: 0  Warnings: 0

18. root@test 11:03:29>INSERT INTO t1 SELECT f1, f2 + 20 FROM t1;
19. Query OK, 40 rows affected (0.00 sec)
20. Records: 40  Duplicates: 0  Warnings: 0

22. root@test 11:03:29>INSERT INTO t1 SELECT f1, f2 + 40 FROM t1;
23. Query OK, 80 rows affected (0.00 sec)
24. Records: 80  Duplicates: 0  Warnings: 0

26. root@test 11:03:29>ANALYZE TABLE t1;
27. +---------+---------+----------+----------+
28. | Table   | Op      | Msg_type | Msg_text |
29. +---------+---------+----------+----------+
30. | test.t1 | analyze | status   | OK       |
31. +---------+---------+----------+----------+
32. 1 row in set (0.00 sec)

34. root@test 11:03:29>EXPLAIN SELECT f1, f2 FROM t1 WHERE f2 > 40;
35. +----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------------+
36. | id | select_type | table | partitions | type  | possible_keys | key     | key_len | ref  | rows | filtered | Extra                                  |
37. +----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------------+
38. |  1 | SIMPLE      | t1    | NULL       | range | PRIMARY       | PRIMARY | 8       | NULL |   53 |   100.00 | Using where; Using index for skip scan |
39. +----+-------------+-------+------------+-------+---------------+---------+---------+------+------+----------+----------------------------------------+
40. 1 row in set, 1 warning (0.00 sec)

也可以从optimizer trace里看到如何选择的skip scan:


1. "skip_scan_range": {
2. "potential_skip_scan_indexes": [
3. {
4. "index": "PRIMARY",
5. "tree_travel_cost": 0.4,
6. "num_groups": 3,
7. "rows": 53,
8. "cost": 10.625
9. }
10. ]
11. },
12. "best_skip_scan_summary": {
13. "type": "skip_scan",
14. "index": "PRIMARY",
15. "key_parts_used_for_access": [
16. "f1",
17. "f2"
18. ],
19. "range": [
20. "40 < f2"
21. ],
22. "chosen": true
23. },

我们从innodb的角度来看看这个SQL是如何执行的，我们知道每个index scan都会走到ha_innobase::index_read来构建search tuple，上述查询的执行步骤：

第一次从Index left side开始scan
第二次使用key(1,40) 扫描index，直到第一个range结束
使用key(1), find_flag =HA_READ_AFTER_KEY, 找到下一个Key值2
使用key(2,40)，扫描Index，直到range结束
使用Key(2)，去找大于2的key值，上例中没有，因此结束扫描

笔者在代码注入了日志，打印search_tuple(dtuple_print())


1. STEP 1: no search_tuple

3. STEP 2:
4. DATA TUPLE: 2 fields;
5. 0: len 4; hex 80000001; asc     ;;
6. 1: len 4; hex 80000028; asc    (;;

8. STEP 3:
9. DATA TUPLE: 1 fields;
10. 0: len 4; hex 80000001; asc     ;;

12. STEP 4:
13. DATA TUPLE: 2 fields;
14. 0: len 4; hex 80000002; asc     ;;
15. 1: len 4; hex 80000028; asc    (;;

17. STEP 5:
18. DATA TUPLE: 1 fields;
19. 0: len 4; hex 80000002; asc     ;;

从上述描述可以看到使用skip-scan的方式避免了全索引扫描，从而提升了性能，尤其是在索引前缀列区分度比较低的时候

条件 skip scan可以通过Hint或者optimizer_switch来控制(skip_scan)，默认是打开的。根据worklog的描述，对于如下query:


1. SELECT A_1,...,A_k, B_1,...,B_m, C
2. FROM T
3. WHERE
4. EQ(A_1,...,A_k)
5. AND RNG(C);

需要满足如下条件才能使用 skip scan:


1. A) Table T has at least one compound index I of the form:
2. I = <A_1,...,A_k, B_1,..., B_m, C ,[D_1,...,D_n]>
3. Key parts A and D may be empty, but B and C must be non-empty.
4. B) Only one table referenced.
5. C) Cannot have group by/select distinct
6. D) Query must reference fields in the index only.
7. E) The predicates on A_1...A_k must be equality predicates and they need
8. to be constants. This includes the 'IN' operator.
9. F) The query must be a conjunctive query.
10. In other words, it is a AND of ORs:
11. (COND1(kp1) OR COND2(kp1)) AND (COND1(kp2) OR ...) AND ...
12. G) There must be a range condition on C.
13. H) Conditions on D columns are allowed. Conditions on D must be in
14. conjunction with range condition on C.

ref: get_best_skip_scan()

当skip scan拥有更低的cost时，会被选择，计算cost的函数是cost_skip_scan()，由于索引统计信息中已经基于不同的前缀列值估算了distinct value的个数(rec_per_key), 可以基于此去预估可能需要读的行数。更具体的可以参考wl#11322中的描述，笔者对此不甚了解，故不做笔墨 ref: cost_skip_scan()

参考

官方文档：Skip Scan Range Access Method WL#11322: SUPPORT LOOSE INDEX RANGE SCANS FOR LOW CARDINALITY Bug#88103 相关代码

原文：http://mysql.taobao.org/monthly/2019/05/06/