PostgreSQL查询优化简介

简介

PostgreSQL查询优化器执行过程

1. 语法分析：生成查询树
2. 语义检查：对SQL表达的语义进行检查
3. 查询优化
   1. 视图重写
   2. 逻辑优化：子查询优化，条件化简，等价谓词重写，连接消除，得到逻辑计划
   3. 物理优化：基于代价优化，得到物理计划。PostgreSQL主要采用动态规划和遗传算法
   4. 非SPJ优化：主要针对分组，排序，去重等操作
4. 查询计划执行

在PostgreSQL中，语法树并不是一棵树状结构的，把关系平面化到一个链表里面。因为，PostgreSQL认为，在这个阶段不清楚表之间如何链接。

重要数据结构

查询语法树

typedef struct Query
{
    //上面还有节点类型，是否存在相关子句
    List	   *cteList;		/* WITH 子句 */
    List	   *rtable;			/* list of range table entries */
    FromExpr   *jointree;       /* table join tree (FROM and WHERE clauses) */
    List	   *targetList;		/* target list (of TargetEntry) */
    List	   *returningList;	/* return-values list (of TargetEntry) */
    List	   *groupClause;	/* a list of SortGroupClause's */
    Node	   *havingQual;		/* qualifications applied to groups */
    List	   *windowClause;	/* 窗口函数子句链表 */
    List	   *distinctClause; /* a list of SortGroupClause's */
    List	   *sortClause;		/* a list of SortGroupClause's */

    Node	   *limitOffset;	/* limit的offset子句 */
    Node	   *limitCount;		/* limit的个数*/
    Node	   *setOperations;	/* 是否为多个SQL UNION/INTERSECT/EXCEPT query */
} Query;

范围表(优化前)

表示被查询的对象，可以是一张表，一个From子句中的子查询，一个连接操作的结果

typedef struct RangeTblEntry
{
    //普通表
    Oid			relid;			/* OID of the relation */
    char		relkind;		/* relation kind (see pg_class.relkind) */
    struct TableSampleClause *tablesample;		/* sampling info, or NULL */

    //子查询
    Query	   *subquery;		/* the sub-query */
    bool		security_barrier;		/* is from security_barrier view?如果是视图展开的子查询，PostgreSQL不做优化 */

    //连接类型
    JoinType	jointype;		/* type of join */
    List	   *joinaliasvars;	/* list of alias-var expansions */
} RangeTblEntry;

关系优化信息(优化过程中)

对应PlannerInfo结构体的两个成员(simple_rel_array和join_rel_list)，是优化阶段的操作对象，具有查询优化的相关信息

typedef struct RelOptInfo
{
    /* all relations included in this RelOptInfo */
    Relids		relids;			/* set of base relids (rangetable indexes) */

    /* 估算结果的行数 */
    double		rows;

    /* materialization information */
    List	   *pathlist;		/* 存放所有可能的路径 */
    List	   *ppilist;		/* ParamPathInfos used in pathlist */
    List	   *partial_pathlist;		/* partial Paths */

    /* 局部最优不一定后面最优，上一层的3个可能的最优结果 */
    struct Path *cheapest_startup_path;
    struct Path *cheapest_total_path;
    struct Path *cheapest_unique_path;
    List	   *cheapest_parameterized_paths;

    //本关系为单表或者join

    /* used by various scans and joins: */
    List	   *baserestrictinfo;		/* RestrictInfo structures (if base
    									 * rel) */
    QualCost	baserestrictcost;		/* cost of evaluating the above */
    List	   *joininfo;		/* RestrictInfo structures for join clauses
    							 * involving this rel */
    bool		has_eclass_joins;		/* T means joininfo is incomplete */
    } RelOptInfo;

计划节点信息

全局查询优化计划的相关信息，存放在PlannerInfo结构体

typedef struct PlannerInfo
{
    Query	   *parse;			/* 开始时的查询计划树 */
    PlannerGlobal *glob;		/* global info for current planner run */
    Index		query_level;	/* 本计划所处的层数*/

    struct PlannerInfo *parent_root;	/* NULL at outermost Query */

    struct RelOptInfo **simple_rel_array;		/* 所有基本表信息 */
    int			simple_rel_array_size;	/* allocated size of array */
    RangeTblEntry **simple_rte_array;	/* rangetable as an array */

    //考虑过连接后生成的新关系
    List	   *join_rel_list;	/* list of join-relation RelOptInfos */
    struct HTAB *join_rel_hash; /* optional hashtable for join relations */

    List	  **join_rel_level; /*结果关系*/
    int			join_cur_level; /* index of list being extended */

} PlannerInfo;

计划节点

代表根据最有路径，生成的物理计划(Plan)

typedef struct Plan
{
    /*
     * estimated execution costs for plan (see costsize.c for more info)
     */
    Cost		startup_cost;	/* cost expended before fetching any tuples */
    Cost		total_cost;		/* total cost (assuming all tuples fetched) */

    /*
     * 估计的元组数和元组宽度
     */
    double		plan_rows;
    int			plan_width;	

    /*
     * Common structural data for all Plan types.
     */
    int			plan_node_id;	/* unique across entire final plan tree */
    List	   *targetlist;		/* target list to be computed at this node */
    List	   *qual;			/* implicitly-ANDed qual conditions */
    struct Plan *lefttree;		/* input plan tree(s) */
    struct Plan *righttree;
    List	   *initPlan;		/* Init Plan nodes (un-correlated expr
    							 * subselects) */
} Plan;

PlannedStmt

优化器结果，保存查询执行计划，范围表相关信息

SelectStmt

语法分析结果

typedef struct SelectStmt
{
    List	   *distinctClause; /* distinct子句*/
    IntoClause *intoClause;		/* target for SELECT INTO */
    List	   *targetList;		/* 投影列子句 */
    List	   *fromClause;		/* From子句，包括join */
    Node	   *whereClause;	/* where子句 */
    List	   *groupClause;	/* GROUP BY clauses */
    Node	   *havingClause;	/* HAVING conditional-expression */
    List	   *windowClause;	/* WINDOW window_name AS (...), ... */

    List	   *sortClause;		/* sort clause (a list of SortBy's) */
    Node	   *limitOffset;	/* # of result tuples to skip */
    Node	   *limitCount;		/* # of result tuples to return */
    List	   *lockingClause;	/* FOR UPDATE (list of LockingClause's) */
} SelectStmt;

结构体关系

1. PlannerInfo是逻辑优化的主要产物，拥有查询树(Query)，关系优化信息，约束条件
2. 路径是物理优化阶段的主要产物，拥有排序键和连接节点
3. PlannerInfo和路径混杂在一起
4. 查询执行计划Plan是所有路径的最小代价生成的

代码入口

planner：主入口函数

PlannedStmt *
planner(Query *parse, int cursorOptions, ParamListInfo boundParams)
{
    PlannedStmt *result;

    if (planner_hook)
    	result = (*planner_hook) (parse, cursorOptions, boundParams);
    else
    	result = standard_planner(parse, cursorOptions, boundParams);
    return result;
}

被主函数调用

standard_planner——标准的查询优化器入口

standard_planner只是查询优化器的外壳，通过调用subquery_planner完成查询优化，通过调用set_plan_references完成清理辅助工作

PlannedStmt *
standard_planner(Query *parse, int cursorOptions, ParamListInfo boundParams)
{
    PlannedStmt *result;//结果
    PlannerGlobal *glob;//查询优化一些所有子查询需要的公共信息
    double		tuple_fraction;//
    PlannerInfo *root;
    RelOptInfo *final_rel;
    Path	   *best_path;
    Plan	   *top_plan;
    ListCell   *lp,
    		   *lr;

···    

    /* primary planning entry point (may recurse for subqueries) */
    root = subquery_planner(glob, parse, NULL,
    						false, tuple_fraction);

    /* Select best Path and turn it into a Plan */
    final_rel = fetch_upper_rel(root, UPPERREL_FINAL, NULL);
    best_path = get_cheapest_fractional_path(final_rel, tuple_fraction);

    top_plan = create_plan(root, best_path);

···        

    /* final cleanup of the plan */
    Assert(glob->finalrtable == NIL);
    Assert(glob->finalrowmarks == NIL);
    Assert(glob->resultRelations == NIL);
    top_plan = set_plan_references(root, top_plan);
    /* ... and the subplans (both regular subplans and initplans) */
    Assert(list_length(glob->subplans) == list_length(glob->subroots));
    forboth(lp, glob->subplans, lr, glob->subroots)
    {
    	Plan	   *subplan = (Plan *) lfirst(lp);
    	PlannerInfo *subroot = (PlannerInfo *) lfirst(lr);

    	lfirst(lp) = set_plan_references(subroot, subplan);
    }

···
    /* build the PlannedStmt result */
    result = makeNode(PlannedStmt);

    result->commandType = parse->commandType;
    result->queryId = parse->queryId;   //parse结果
    result->planTree = top_plan;    //查询计划
    result->rtable = glob->finalrtable; //范围表
    result->resultRelations = glob->resultRelations;    //

    return result;
}

1. subquery_planner返回逻辑优化和物理优化结果root(PlannerInfo *)
2. create_plan根据最优路径，和PlannerInfo生成物理执行计划Plan
3. set_plan_references对执行计划部分调整和清理

subquery_planner生成(子)查询执行计划的函数

subquery_planner分为两步。第一步是逻辑优化；第二步是物理优化

//传入glob，parse,typle_fraction，parent_root最开始为NULL
PlannerInfo *
subquery_planner(PlannerGlobal *glob, Query *parse,
				 PlannerInfo *parent_root,
				 bool hasRecursion, double tuple_fraction)
{
    PlannerInfo *root;
    List	   *newWithCheckOptions;
    List	   *newHaving;
    bool		hasOuterJoins;
    RelOptInfo *final_rel;
    ListCell   *l;

    /* 为当前子查询创建PlannerInfo */
    root = makeNode(PlannerInfo);
    root->parse = parse;
    root->glob = glob;
    root->query_level = parent_root ? parent_root->query_level + 1 : 1;
    root->parent_root = parent_root;
    root->hasRecursion = hasRecursion;
    if (hasRecursion)
    	root->wt_param_id = SS_assign_special_param(root);
    else
    	root->wt_param_id = -1;
    root->non_recursive_path = NULL;

    /*
     * Look for ANY and EXISTS SubLinks in WHERE and JOIN/ON clauses, and try
     * to transform them into joins.  Note that this step does not descend
     * into subqueries; if we pull up any subqueries below, their SubLinks are
     * processed just before pulling them up.
     *子连接：在where和join子句含有ANY和EXISTS
     */
    if (parse->hasSubLinks)
    	pull_up_sublinks(root);

    //上拉子查询
    pull_up_subqueries(root);

    //子查询合并
    if (parse->setOperations)
    	flatten_simple_union_all(root);

    //上拉子查询后处理继承关系
    preprocess_rowmarks(root);
    expand_inherited_tables(root);

    //条件化简
    preprocess_expression

    //合并having子句到where子句，如果having里面含有聚集函数，易失函数，子查询不能合并    

    //消除外连接
    if (hasOuterJoins)
    	reduce_outer_joins(root);

    /*
     * Do the main planning.  If we have an inherited target relation, that
     * needs special processing, else go straight to grouping_planner.
     */
    if (parse->resultRelation &&
    	rt_fetch(parse->resultRelation, parse->rtable)->inh)
    	//含有继承关系的物理优化
    	inheritance_planner(root);
    else
        //物理优化
    	grouping_planner(root, false, tuple_fraction);

    set_cheapest(final_rel);

    return root;
}

1. 逻辑优化
   1. 处理CTE表达式（ss_process_ctes）
   2. 上拉子连接
   3. 上拉子查询
   4. Union all处理：flatten_simple_union_all
   5. 处理for update(row lock)：preprocess_rowmark
   6. 继承表处理(expand_inherited_tables)
   7. 处理目标列(prepocess_expression)
   8. 处理withCheckOptions：prepocess_expression
   9. 处理return 表达式，window子句，limit off子句：prepocess_expression
   10. 合并having到where子句
   11. 消除外连接
2. 物理优化：生成本成查询PlanInfo的三条最优路径，返回给上层

整体流程

• subquery_planner
  • 处理CTE表达式
  • 上拉子链接(去除in, some ,exist)
  • 上拉子查询
  • 预处理表达式(and/or， 计算明显的结果， 处理不能上拉的子连接)
  • 消除外连接
  • grouping_planer
    • 处理集合(生成子查询)
    • 处理非集合，调整order, group, target list之句中的顺序
    • query_planner
      • 构建基本表的RelOptInfo
      • 选择下推
      • 投影下推
      • 推导隐含表达式
      • 生成pathkey
      • make_one_rel(通过数据直方图计算选择率)
        • 处理单表最优的查询方式
        • 处理两表join最优的方式(判定是否special join)
        • 动态规划或者遗传算法构建多表Join
    • 获取cheatest_path，再加上其他子句