卖文章的网站源码,公司名字大全列表,wordpress迁移到jekyll,手机平台网站系统简介#xff1a; 本文基于MySQL 8.0.25源码进行分析和总结。这里MySQL Server层指的是MySQL的优化器、执行器部分。我们对MySQL的理解还建立在5.6和5.7版本的理解之上#xff0c;更多的是对比PostgreSQL或者传统数据库。然而从MySQL 8.0开始#xff0c;持续每三个月的迭代和…简介 本文基于MySQL 8.0.25源码进行分析和总结。这里MySQL Server层指的是MySQL的优化器、执行器部分。我们对MySQL的理解还建立在5.6和5.7版本的理解之上更多的是对比PostgreSQL或者传统数据库。然而从MySQL 8.0开始持续每三个月的迭代和重构工作使得MySQL Server层的整体架构有了质的飞越。下面来看下MySQL最新的架构。 作者 | 道客 来源 | 阿里技术公众号
一 背景和架构
本文基于MySQL 8.0.25源码进行分析和总结。这里MySQL Server层指的是MySQL的优化器、执行器部分。我们对MySQL的理解还建立在5.6和5.7版本的理解之上更多的是对比PostgreSQL或者传统数据库。然而从MySQL 8.0开始持续每三个月的迭代和重构工作使得MySQL Server层的整体架构有了质的飞越。下面来看下MySQL最新的架构。 我们可以看到最新的MySQL的分层架构和其他数据库并没有太大的区别另外值得一提的是从图中可以看出MySQL现在更多的加强InnoDB、NDB集群和RAPID(HeatWave clusters)内存集群架构的演进。下面我们就看下具体细节我们这次不随着官方的Feature实现和重构顺序进行理解本文更偏向于从优化器、执行器的流程角度来演进。
二 MySQL 解析器Parser
首先从Parser开始官方MySQL 8.0使用Bison进行了重写生成Parser Tree同时Parser Tree会contextualize生成MySQL抽象语法树Abstract Syntax Tree。 MySQL抽象语法树和其他数据库有些不同是由比较让人拗口的SELECT_LEX_UNIT/SELECT_LEX类交替构成的然而这两个结构在最新的版本中已经重命名成标准的SELECT_LEX - Query_block和SELECT_LEX_UNIT - Query_expression。Query_block是代表查询块而Query_expression是包含多个查询块的查询表达式包括UNION AND/OR的查询块如SELECT FROM t1 union SELECT FROM t2或者有多Level的ORDER BY/LIMIT (如SELECT * FROM t1 ORDER BY a LIMIT 10) ORDER BY b LIMIT 5。
例如来看一个复杂的嵌套查询 (SELECT *FROM ttt1)
UNION ALL(SELECT *FROM(SELECT *FROM ttt2) AS a,(SELECT *FROM ttt3UNION ALL SELECT *FROM ttt4) AS b)
在MySQL中就可以用下面方式表达 经过解析和转换后的语法树仍然建立在Query_block和Query_expression的框架下只不过有些LEVEL的query block被消除或者合并了这里不再详细展开。
三 MySQL prepare/rewrite阶段
接下来我们要经过resolve和transformation过程Query_expression::prepare-Query_block::prepare这个过程包括按功能分而非完全按照执行顺序
1 Setup and Fix
setup_tablesSet up table leaves in the query block based on list of tables.resolve_placeholder_tables/merge_derived/setup_table_function/setup_materialized_derivedResolve derived table, view or table function references in query block.setup_natural_join_row_typesCompute and store the row types of the top-most NATURAL/USING joins.setup_wildExpand all * in list of expressions with the matching column references.setup_base_ref_itemsSet query_blocks base_ref_items.setup_fieldsCheck that all given fields exists and fill struct with current data.setup_condsResolve WHERE condition and join conditions.setup_groupResolve and set up the GROUP BY list.m_having_cond-fix_fieldsSetup the HAVING clause.resolve_rollupResolve items in SELECT list and ORDER BY list for rollup processing.resolve_rollup_itemResolve an item (and its tree) for rollup processing by replacing items matching grouped expressions with Item_rollup_group_items and updating properties (m_nullable, PROP_ROLLUP_FIELD). Also check any GROUPING function for incorrect column.setup_orderSet up the ORDER BY clause.resolve_limitsResolve OFFSET and LIMIT clauses.Window::setup_windows1Set up windows after setup_order() and before setup_order_final().setup_order_finalDo final setup of ORDER BY clause, after the query block is fully resolved.setup_ftfuncsSetup full-text functions after resolving HAVING.resolve_rollup_wfs : Replace group by field references inside window functions with references in the presence of ROLLUP.
2 Transformation
remove_redundant_subquery_clause : Permanently remove redundant parts from the query if 1) This is a subquery 2) Not normalizing a view. Removal should take place when a query involving a view is optimized, not when the view is created.remove_base_optionsRemove SELECT_DISTINCT options from a query block if can skip distinct. resolve_subquery : Resolve predicate involving subquery, perform early unconditional subquery transformations. Convert subquery predicate into semi-join, orMark the subquery for execution using materialization, orPerform IN-EXISTS transformation, orPerform more/less ALL/ANY - MIN/MAX rewriteSubstitute trivial scalar-context subquery with its valuetransform_scalar_subqueries_to_join_with_derivedTransform eligible scalar subqueries to derived tables.flatten_subqueriesConvert semi-join subquery predicates into semi-join join nests. Convert candidate subquery predicates into semi-join join nests. This transformation is performed once in query lifetime and is irreversible. apply_local_transforms : delete_unused_merged_columns : If query block contains one or more merged derived tables/views, walk through lists of columns in select lists and remove unused columns.simplify_joinsConvert all outer joins to inner joins if possibleprune_partitionsPerform partition pruning for a given table and condition.push_conditions_to_derived_tablesPushing conditions down to derived tables must be done after validity checks of grouped queries done by apply_local_transforms();Window::eliminate_unused_objectsEliminate unused window definitions, redundant sorts etc.
这里节省篇幅我们只举例关注下和top_join_list相关的simple_joins这个函数的作用对于Query_block中嵌套join的简化过程。 3 对比PostgreSQL
为了更清晰的理解标准数据库的做法我们这里引用了PostgreSQL的这三个过程
Parser
下图首先Parser把SQL语句生成parse tree。
testdb# SELECT id, data FROM tbl_a WHERE id 300 ORDER BY data;Analyzer/Analyser
下图展示了PostgreSQL的analyzer/analyser如何将parse tree通过语义分析后生成query tree。 Rewriter
Rewriter会根据规则系统中的规则把query tree进行转换改写。
sampledb# CREATE VIEW employees_list
sampledb-# AS SELECT e.id, e.name, d.name AS department
sampledb-# FROM employees AS e, departments AS d WHERE e.department_id d.id;下图的例子就是一个包含view的query tree如何展开成新的query tree。
sampledb# SELECT * FROM employees_list;四 MySQL Optimize和Planning阶段
接下来我们进入了逻辑计划生成物理计划的过程本文还是注重于结构的解析而不去介绍生成的细节MySQL过去在8.0.22之前主要依赖的结构就是JOIN和QEP_TAB。JOIN是与之对应的每个Query_block而QEP_TAB对应的每个Query_block涉及到的具体“表”的顺序、方法和执行计划。然而在8.0.22之后新的基于Hypergraph的优化器算法成功的抛弃了QEP_TAB结构来表达左深树的执行计划而直接使用HyperNode/HyperEdge的图来表示执行计划。 举例可以看到数据结构表达的left deep tree和超图结构表达的bushy tree对应的不同计划展现
| - Inner hash join (no condition) (cost1.40 rows1)- Table scan on R4 (cost0.35 rows1)- Hash- Inner hash join (no condition) (cost1.05 rows1)- Table scan on R3 (cost0.35 rows1)- Hash- Inner hash join (no condition) (cost0.70 rows1)- Table scan on R2 (cost0.35 rows1)- Hash- Table scan on R1 (cost0.35 rows1)| - Nested loop inner join (cost0.55..0.55 rows0)- Nested loop inner join (cost0.50..0.50 rows0)- Table scan on R4 (cost0.25..0.25 rows1)- Filter: (R4.c1 R3.c1) (cost0.35..0.35 rows0)- Table scan on R3 (cost0.25..0.25 rows1)- Nested loop inner join (cost0.50..0.50 rows0)- Table scan on R2 (cost0.25..0.25 rows1)- Filter: (R2.c1 R1.c1) (cost0.35..0.35 rows0)- Table scan on R1 (cost0.25..0.25 rows1)
MySQL8.0.2x为了更好的兼容两种优化器引入了新的类AccessPath可以认为这是MySQL为了解耦执行器和不同优化器抽象出来的Plan Tree。 1 老优化器的入口
老优化器仍然走JOIN::optimize来把query block转换成query execution plan (QEP)。
这个阶段仍然做一些逻辑的重写工作这个阶段的转换可以理解为基于cost-based优化前做准备详细步骤如下 Logical transformations optimize_derived : Optimize the query expression representing a derived table/view.optimize_cond : Equality/constant propagation.prune_table_partitions : Partition pruning.optimize_aggregated_query : COUNT(*), MIN(), MAX() constant substitution in case of implicit grouping.substitute_gc : ORDER BY optimization, substitute all expressions in the WHERE condition and ORDER/GROUP lists that match generated columns (GC) expressions with GC fields, if any. Perform cost-based optimization of table order and access path selection. JOIN::make_join_plan() : Set up join order and initial access paths. Post-join order optimization substitute_for_best_equal_field : Create optimal table conditions from the where clause and the join conditions.make_join_query_block : Inject outer-join guarding conditions.Adjust data access methods after determining table condition (several times).optimize_distinct_group_order : Optimize ORDER BY/DISTINCT.optimize_fts_query : Perform FULLTEXT search before all regular searches.remove_eq_conds : Removes const and eq items. Returns the new item, or nullptr if no condition.replace_index_subquery/create_access_paths_for_index_subquery : See if this subquery can be evaluated with subselect_indexsubquery_engine.setup_join_buffering : Check whether join cache could be used. Code generation alloc_qep(tables) : Create QEP_TAB array.test_skip_sort : Try to optimize away sorting/distinct.make_join_readinfo : Plan refinement stage: do various setup things for the executor.make_tmp_tables_info : Setup temporary table usage for grouping and/or sorting.push_to_engines : Push (parts of) the query execution down to the storage engines if they can provide faster execution of the query, or part of it.create_access_paths : generated ACCESS_PATH.
2 新优化器的入口
新优化器默认不打开必须通过set optimizer_switchhypergraph_optimizeron; 来打开。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。
举例看下PlanAccessPath和SQL的关系 最后生成Iterator执行器框架需要的Iterator执行载体AccessPath和Iterator是一对一的关系Access paths are a query planning structure that correspond 1:1 to iterators。 Query_expression::m_root_iterator CreateIteratorFromAccessPath(......)unique_ptr_destroy_onlyRowIterator CreateIteratorFromAccessPath(THD *thd, AccessPath *path, JOIN *join, bool eligible_for_batch_mode) {
......switch (path-type) {case AccessPath::TABLE_SCAN: {const auto param path-table_scan();iterator NewIteratorTableScanIterator(thd, param.table, path-num_output_rows, examined_rows);break;}case AccessPath::INDEX_SCAN: {const auto param path-index_scan();if (param.reverse) {iterator NewIteratorIndexScanIteratortrue(thd, param.table, param.idx, param.use_order, path-num_output_rows,examined_rows);} else {iterator NewIteratorIndexScanIteratorfalse(thd, param.table, param.idx, param.use_order, path-num_output_rows,examined_rows);}break;}case AccessPath::REF: {
......
}
3 对比PostgreSQL
testdb# EXPLAIN SELECT * FROM tbl_a WHERE id 300 ORDER BY data;QUERY PLAN
---------------------------------------------------------------Sort (cost182.34..183.09 rows300 width8)Sort Key: data- Seq Scan on tbl_a (cost0.00..170.00 rows300 width8)Filter: (id 300)
(4 rows) 五 总结
本文主要focus在MySQL最新版本官方的源码上重点分析了官方的重构在多阶段和各阶段结构上的变化和联系更多的是为了让大家了解一个全新的MySQL的发展。
原文链接
本文为阿里云原创内容未经允许不得转载。
