我的书签
添加书签
移除书签MySQL · 源码解析 · MySQL 8.0.23 Hypergraph Join Optimizer代码详解
MySQL JOIN简介
MySQL本身没有常规意义上的执行计划，一般情况就是通过JOIN和QEP_TAB这两个结构组成。QEP_TAB 的全称是Query Execution Plan Table，这个“Table“可以是物理表、内存表、常量表、子查询的结果表等等。作为整个单独JOIN执行计划载体之前还承担着整个执行路径的调用和流转，但是从8.0.20后，全面的生成了独立的Iterator执行器引擎模式。在8.0.22中，又引入了AccessPath概念，真正的生成了独立的执行计划，从而进一步做到了优化过程到树型执行计划，最后到Iterator载体在执行引擎中的执行。
MySQL原始的Join都是依赖于QEP_TAB列表，因为原来MySQL并不支持其他形态的Join结构，只支持左深树，那很容易直接使用数组来表示就可以了。优化器在生成执行计划只需要在QEP_TAB上增加JOIN的属性op_type，就可以递归去使用不同的Join方法和表访问方式了。
1. // Operation between the previous QEP_TAB and this one.
2. enum enum_op_type {
3. // Regular nested loop.
4. OT_NONE,
6. // Aggregate (GROUP BY).
7. OT_AGGREGATE,
9. // Various temporary table operations, used at the end of the join.
10. OT_MATERIALIZE,
11. OT_AGGREGATE_THEN_MATERIALIZE,
12. OT_AGGREGATE_INTO_TMP_TABLE,
13. OT_WINDOWING_FUNCTION,
15. // Block-nested loop (rewritten to hash join).
16. OT_BNL,
18. // Batch key access.
19. OT_BKA
20. } op_type = OT_NONE;
Hypergraph Join Optimizer
官方共分了11个Patch来提交对于Join优化器的增强，当然其中包含了对优化器和执行器分离更进一步重构，我们先来看看官方是怎么提交这样的重大重构的。
[Basic] 动态规划查询超图算法（DPhyp-Hypergraph partitioning algorithm） 官方首先实现了基于DPhyp的动态规划查询超图算法，论文可以搜索《Dynamic Programming Strikes Back》。数据库中关于Join ordering算法有很多，引用2，3中的作者已经做了详尽的解释。我这里只做简单的介绍。
每一个Query，都可以定义为一个无向Query Graph，包括查询中的所有关系R1，R2，…，Rn作为节点；连接谓词表达式作为边，如a1 = a2 （a1 ∈ Ri，a2 ∈ Rj）；连接谓词中包含常量会形成自边（self-edge），如a1 = const (a1 ∈ Ri)；大部分的自边在Join算法里是不考虑的，因为它会被下推下图。例如对于 select * from Student s, Attend a, Lecture l, Professor p where s.sno = a.asno and a.alno = l.lno and l.lpno = p.pno and p.pname = ‘Larson’，有如下Query Graph结构：
对于Join Tree，一般会有以下几种：left-deep tree、right-deep tree、zigzag tree和bushy tree。前三种是属于线性Join tree。MySQL之前采取左深树，为了考虑更好的支持Hash Join和NestLoop Join的选择，现在开始考虑Bushy Tree了。为了避免任何时候的笛卡尔积Join，线性Join的Join ordering算法通常很简单。那么为什么要引入复杂的Bushy Tree。假设定义Query(R1, R2, R3)有如下属性，y |R1| = 10, |R2| = 20, |R3| = 20, |R4| = 10, f1,2 = 0.01, f2,3 = 0.5, f3,4 = 0.01。||代表行数，fn,m代表Rn和Rm的选择率，可以看到Bushy Tree有更好的执行效率。
不过遗憾的是，Bushy Tree的搜索可能性非常大：
因此，原始左深树使用的Greedy Heuristics算法，在Bushy Tree下，计算Join Ordering通常使用动态规划算法（DPccp和DPhyp）。
DPccp的算法如下：
但是DPccp有很多限制：复杂谓词，涉及到多个表（R1，R2，R3）做为连接，例如：R1.a + R2.b + R3.c = R4.d + R5.e + R6.f ；只支持inner joins；因此引入了新的基于Hypergraph的算法DPhyp。
1. select *
2. from R1 r1, R2 r2, R3 r3,
3. R4 r4, R5 r5, R6 r6
4. where r1.a=r2.a and r2.b=r3.c and
5. r4.d=r5.d and r5.e=r6.e and
6. abs(r1.f + r3.f )
7. = abs(r4.g + r6.g)
介绍算法先介绍下基本概念超图（hypergraph）相比普通的图，其特点是图中的节点是一个集合，称为超节点（hypernode），图中的边所连接的是超节点，即连接两个集合。这类边称为超边（hyperedge）。超图就是由超节点和超边作为最基本元素而构成的。有了超图那么上面的Join Graph可以变成：
由于使用DPccp和Top-Down Partition Search，不能够解决outer join，antijoin的不能自由重排的算法。
MySQL目前采用Bitmap（64bit）来表示，假设Join table个数不会超过61个，看下它的定义
1. +struct Hyperedge {
2. + // The endpoints (hypernodes) of this hyperedge. See the comment about
3. + // duplicated edges in Node.
4. + //
5. + // left and right may not overlap, and both must have at least one bit set.
6. + NodeMap left;
7. + NodeMap right;
8. +};
9. +
10. +struct Hypergraph {
11. + std::vector<Node> nodes; // Maximum 8*sizeof(NodeMap) elements.
12. + std::vector<Hyperedge> edges;
13. +
14. + void AddNode();
15. + void AddEdge(NodeMap left, NodeMap right);
16. +};
基本算法流程如下：
- 找到一个图中种子节点Ri
- 不断增加i去找hyperedges，不考虑不连接的和已经处理过的。
- 对于每一个连通子图subgraph (csg)，再重复1和2步骤，找出一个仍然可以连通子图(complement, cmp)，然后连接这个图的cmp成为更大的连通子图(csg-cmp-pair).
- 当找到一个csg-cmp-pair，就形成一个可以进行估算的subjoin。
感兴趣可以阅读相应的论文和MySQL的代码（sql/join_optimizer）。
QEP_TAB和执行器Iterator解藕，重新来设置InnoDB row buffer
众所周知，QEB_TAB结构上承载了很多信息，除了上面表访问和Join方法的信息之外，还有InnoDB row buffer、表访问的优化访问方式（ref/range/loose scan/first match/materialize）、附加属性（having/distinct/sort/icp/lateral derived/mrr/cte）、基本物理表结构TABLE_LIST等。作为删除QEP_TAB的基础，首先先做了和执行器的解藕工作，Iterator和QEP_TAB分离。
1. class TableScanIterator final : public TableRowIterator {
2. public:
3. - // Accepts nullptr for qep_tab; qep_tab is used only for setting up record
4. - // buffers.
5. - //
6. - // The pushed condition can be nullptr.
7. + // “expected_rows” is used for scaling the record buffer.
8. + // If zero or less, no record buffer will be set up.
9. //
10. // "examined_rows", if not nullptr, is incremented for each successful Read().
11. - TableScanIterator(THD *thd, TABLE *table, QEP_TAB *qep_tab,
12. + TableScanIterator(THD *thd, TABLE *table, double expected_rows,
13. ha_rows *examined_rows);
接下来解藕
1. -static bool init_index_and_record_buffer(const QEP_TAB *qep_tab, handler *file,
2. +static bool init_index(TABLE *table, handler *file, uint idx, bool sorted) {
4. -bool set_record_buffer(const QEP_TAB *tab);
5. +bool set_record_buffer(TABLE *table, double expected_rows_to_fetch);
7. =>
9. - return init_index_and_record_buffer(m_qep_tab, m_qep_tab->table()->file,
10. - m_ref->key, m_use_order);
11. + if (table()->file->inited) return false;
12. + if (init_index(table(), table()->file, m_ref->key, m_use_order)) {
13. + return true;
14. + }
15. + return set_record_buffer(table(), m_expected_rows);
实现CostingReceiver和转化查询块select_lex成为超图hypergraph
MySQL 8.0.23提供了支持hypergraph的优化器模型的第一个原型版本，通过set optimizer_switch=”hypergraph_optimizer=on”;来打开，主要和原有的优化器区别在于：
- 不再局限于左深树的执行计划
- 用DPhyp动态规划算法代替了强力算和启发式的剪枝方式，减少了搜索空间，当然还有一些限制
- Hash join成为主要的选择方式
- 直接和AccessPath互通，而非直接生成Iterators
主要通过FindBestQueryPlan函数来实现，逻辑如下：
- 先判断是否属于新优化器可以支持的Query语法（CheckSupportedQuery），不支持的直接返回错误ER_HYPERGRAPH_NOT_SUPPORTED_YET
- 转化top_join_list变成JoinHypergraph结构。由于Hypergraph是比较独立的算法层面的实现，JoinHypergraph结构用来更好的把数据库的结构包装到Hypergraph的edges和nodes的概念上的。
- 通过EnumerateAllConnectedPartitions实现论文中的DPhyp算法
- CostingReceiver类包含了过去JOIN planning的主要逻辑，包括根据cost选择相应的访问路径，根据DPhyp生成的子计划进行评估，保留cost最小的子计划。
- 得到root_path后，接下来处理group/agg/having/sort/limit的。对于Group by操作，目前Hypergraph使用sorting first + streaming aggregation的方式。
FindBestQueryPlan最终返回确定的执行计划root_path后，通过CreateIteratorFromAccessPath函数生成对应的执行Iterator树，在Iterator执行器中执行。
举例说明：
1. 两个连通子图
3. root:test> explain format=tree select * from t1,t2,t3,t4 where t2.f2 = t1.a and t1.a = t3.a;
4. +------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
5. | EXPLAIN |
6. +------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
7. | -> Hash cartesian product (no condition) (cost=1.83 rows=2)
8. -> Inner hash join (t2.f2 = t1.a) (cost=1.55 rows=2)
9. -> Table scan on t2 (cost=0.25 rows=2)
10. -> Hash
11. -> Inner hash join (t1.a = t3.a) (cost=1.27 rows=1)
12. -> Table scan on t1 (cost=1.00 rows=1)
13. -> Hash
14. -> Table scan on t3 (cost=0.25 rows=1)
15. -> Hash
16. -> Table scan on t4 (cost=0.25 rows=1)
17. |
19. 一个连通子图
21. root:test> explain format=tree select * from t1,t2,t3,t4 where t2.f2 = t1.a and t1.a = t3.a and t2.f2 = t4.pk and t1.a = t4.pk;
22. +----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
23. | EXPLAIN |
24. +----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
25. | -> Inner hash join (t2.f2 = t4.pk), (t1.a = t4.pk) (cost=1.83 rows=2)
26. -> Inner hash join (t2.f2 = t1.a) (cost=1.55 rows=2)
27. -> Table scan on t2 (cost=0.25 rows=2)
28. -> Hash
29. -> Inner hash join (t1.a = t3.a) (cost=1.27 rows=1)
30. -> Table scan on t1 (cost=1.00 rows=1)
31. -> Hash
32. -> Table scan on t3 (cost=0.25 rows=1)
33. -> Hash
34. -> Table scan on t4 (cost=0.25 rows=1)
35. |
通过打开opt_trace来看下整个过程
1. {
2. "steps": [
3. {
4. "join_preparation": {
5. "select#": 1,
6. "steps": [
7. {
8. "expanded_query": "/* select#1 */ select `t1`.`a` AS `a`,`t2`.`f2` AS `f2`,`t3`.`a` AS `a`,`t3`.`b` AS `b`,`t4`.`pk` AS `pk`,`t4`.`col_int_nokey` AS `col_int_nokey`,`t4`.`col_int_key` AS `col_int_key`,`t4`.`col_varchar_nokey` AS `col_varchar_nokey` from `t1` join `t2` join `t3` join `t4` where ((`t2`.`f2` = `t1`.`a`) and (`t2`.`f2` = `t4`.`col_int_key`))"
9. }
10. ]
11. }
12. },
13. {
14. "join_optimization": {
15. "select#": 1,
16. "steps": [
17. {
18. "condition_processing": {
19. "condition": "WHERE",
20. "original_condition": "((`t2`.`f2` = `t1`.`a`) and (`t2`.`f2` = `t4`.`col_int_key`))",
21. "steps": [
22. {
23. "transformation": "equality_propagation",
24. "resulting_condition": "((`t2`.`f2` = `t1`.`a`) and (`t2`.`f2` = `t4`.`col_int_key`))"
25. },
26. {
27. "transformation": "trivial_condition_removal",
28. "resulting_condition": "((`t2`.`f2` = `t1`.`a`) and (`t2`.`f2` = `t4`.`col_int_key`))"
29. }
30. ]
31. }
32. },
33. {
34. "substitute_generated_columns": {
35. }
36. },
37. {
38. "join_optimizer": [
39. "Join list after simplification:",
40. "* t4 join_type=inner",
41. "* t3 join_type=inner",
42. "* t2 join_type=inner",
43. "* t1 join_type=inner",
44. "",
45. "Made this relational tree; WHERE condition is ((t2.f2 = t1.a) and (t2.f2 = t4.col_int_key)):",
46. "* Inner join (no join conditions)",
47. " * Inner join (no join conditions)",
48. " * Inner join (no join conditions)",
49. " * t1",
50. " * t2",
51. " * t3",
52. " * t4",
53. "",
54. "After pushdown; remaining WHERE conditions are (none):",
55. "* Inner join (equijoin condition = (t2.f2 = t4.col_int_key))",
56. " * Cartesian product",
57. " * Inner join (equijoin condition = (t2.f2 = t1.a))",
58. " * t1",
59. " * t2",
60. " * t3",
61. " * t4",
62. "",
63. "Selectivity of join (t2.f2 = t1.a):",
64. " - found an index in t1.a for (t2.f2 = t1.a), selectivity = 1.000",
65. "Selectivity of join [cartesian product]:",
66. "Selectivity of join (t2.f2 = t4.col_int_key):",
67. " - found an index in t4.col_int_key for (t2.f2 = t4.col_int_key), selectivity = 1.000",
68. "",
69. "Constructed hypergraph:",
70. "digraph G { # 3 edges",
71. " t1 -> t2 [label=\"(t2.f2 = t1.a)\"]",
72. " e2 [shape=circle,width=.001,height=.001,label=\"\"]",
73. " t1 -> e2 [arrowhead=none,label=\"\"]",
74. " t2 -> e2 [arrowhead=none,label=\"\"]",
75. " e2 -> t3 [label=\"[cartesian product]\"]",
76. " t2 -> t4 [label=\"(t2.f2 = t4.col_int_key)\"]",
77. "}",
78. "",
79. "Enumerating subplans:",
80. "Found node t4 [rows=1, cost=0.2]",
81. "Found node t3 [rows=1, cost=0.2]",
82. "Found node t2 [rows=2, cost=0.2]",
83. "Found sets {t2} and {t4}, connected by condition (t2.f2 = t4.col_int_key) [rows=2, cost=0.5]",
84. " - first alternative for this join, keeping",
85. "Found node t1 [rows=1, cost=1.0]",
86. "Found sets {t2} and {t1}, connected by condition (t2.f2 = t1.a) [rows=2, cost=1.3]",
87. " - first alternative for this join, keeping",
88. "Found sets {t2,t4} and {t1}, connected by condition (t2.f2 = t1.a) [rows=2, cost=1.6]",
89. " - first alternative for this join, keeping",
90. "Found sets {t1,t2} and {t4}, connected by condition (t2.f2 = t4.col_int_key) [rows=2, cost=1.6]",
91. " - more expensive than old cost 1.6, discarding",
92. "Found sets {t1,t2} and {t3}, connected by condition [cartesian product] [rows=2, cost=1.6]",
93. " - first alternative for this join, keeping",
94. "Found sets {t1,t2,t3} and {t4}, connected by condition (t2.f2 = t4.col_int_key) [rows=2, cost=1.8]",
95. " - first alternative for this join, keeping",
96. "Found sets {t1,t2,t4} and {t3}, connected by condition [cartesian product] [rows=2, cost=1.8]",
97. " - more expensive than old cost 1.8, discarding",
98. "",
99. "Enumerated 9 subplans."
100. ]
101. }
102. ]
103. }
104. }
105. ]
106. }
可以看到官方MySQL还提供了查看Graph图的Json格式可视化脚本，我们可以通过online graph看到连通图：
参考资料：
- Dynamic Programming Strikes Back
- Counter Strike: Generic Top-Down Join Enumeration for Hypergraphs
- MySQL 8.0 新的火山模型执行器
- SIGMOD08 Join Ordering:应对复杂谓词和非内连接的场景
