MySQL 5.7的Optimizer跟踪解析

                          MySQL 5.7的Optimizer跟踪解析

    在Mysql 5.7以后引入了强大的optmizer跟踪,这样可以帮助我们在分析sql执行计划时了解
mysql为什么会选择某种执行带来很大方便.对于分析和调整sql语句性能有很大帮助

 测试环境

mysql> select version();
+------------+
| version()  |
+------------+
| 5.7.10-log |
+------------+

mysql> create table t1 (a int auto_increment primary key, b int,c int);
Query OK, 0 rows affected (0.01 sec)

mysql> alter table t1 add key ix_b (b);

插入一些测试数据

打开优化器跟踪

mysql> set optimizer_trace='enabled=on'

执行语句

mysql> select * from t1 where b=2

从该语句来考察,我们知道该语句有全表扫描和扫描索引再扫描表数据的两种方式.
我们来看看mysql的优化器是怎么认为的

查看跟踪数据.
mysql> select * from information_schema.OPTIMIZER_TRACE

该视图只能是在当前跟踪的会话中看到,所以必须要和执行的sql在同一个会话

 跟踪的结果接近有两百行,我们一步一步来看mysql的优化器是怎么评估的

{
  "steps": [
    {
      "join_preparation": {
        "select#": 1,
        "steps": [
          {
            "expanded_query": "/* select#1 */ select `t1`.`a` AS `a`,`t1`.`b` AS `b`,`t1`.`c` AS `c` from `t1` where

(`t1`.`b` = 2)"

 ---join_preparation,这一步也称为查询转换,可以看到我们执行的select *,而mysql会把所有的栏位都解析出来,最终执行的sql是这个
样子,在这一步还会做一些视图转换,子查询转换等一些工作.

          }
        ]
      }
    },
    {
      "join_optimization": {

  --查询优化阶段

        "select#": 1,
        "steps": [
          {
            "condition_processing": {
              "condition": "WHERE",
              "original_condition": "(`t1`.`b` = 2)",
              "steps": [
                {
                  "transformation": "equality_propagation",
                  "resulting_condition": "multiple equal(2, `t1`.`b`)"
                },
                {
                  "transformation": "constant_propagation",
                  "resulting_condition": "multiple equal(2, `t1`.`b`)"
                },
                {
                  "transformation": "trivial_condition_removal",
                  "resulting_condition": "multiple equal(2, `t1`.`b`)"
                }
              ]
            }
          },
         
         --条件处理,相等传播,常量传播,条件移除(主要是对一些无关的条件进行消除）

          {
            "substitute_generated_columns": {
            }
          },
              --替代产生的栏位

          {
            "table_dependencies": [
              {
                "table": "`t1`",
                "row_may_be_null": false,
                "map_bit": 0,
                "depends_on_map_bits": [
                ]
              }
            ]
          },
          {
            "ref_optimizer_key_uses": [
              {
                "table": "`t1`",
                "field": "b",
                "equals": "2",
                "null_rejecting": false
              }
            ]
          },
          {
            "rows_estimation": [
         --最重的一步,优化器评估可选的执行方式,
              {
                "table": "`t1`",
                "range_analysis": {
                  "table_scan": {
                    "rows": 7,
                    "cost": 4.5
               --全表扫描,行,成本
                  },
                  "potential_range_indexes": [
                    {
                      "index": "PRIMARY",
                      "usable": false,
                      "cause": "not_applicable"
                    --主健扫描,但是不可用,所以没有被选择
                    },
                    {
                      "index": "ix_b",
                      "usable": true,
                      "key_parts": [
                        "b",
                        "a"
                      ]
                    }
                  ],
                  --索引ix_b被选择

                  "setup_range_conditions": [
                  ],
                  "group_index_range": {
                    "chosen": false,
                    "cause": "not_group_by_or_distinct"
                  },
                  "analyzing_range_alternatives": {
                    "range_scan_alternatives": [
                      {
                        "index": "ix_b",
                        "ranges": [
                          "2 <= b <= 2"
                        ],
                        "index_dives_for_eq_ranges": true,

                      --优化器发现索引使用的条件是相等条件

                        "rowid_ordered": true,
                        "using_mrr": false,
                        "index_only": false,
                        "rows": 2,
                        "cost": 3.41,
                        "chosen": true
                      }
                    ],
                    "analyzing_roworder_intersect": {
                      "usable": false,
                      "cause": "too_few_roworder_scans"
                    }
                  },
                  "chosen_range_access_summary": {
                    "range_access_plan": {
                      "type": "range_scan",
                      "index": "ix_b",
                      "rows": 2,
                      "ranges": [
                        "2 <= b <= 2"
                      ]
                    },
                    "rows_for_plan": 2,
                    "cost_for_plan": 3.41,
                    "chosen": true

            --在这个解段发现可以使用range的方式扫描,计算出行和成本等信息
                  }
                }
              }
            ]
          },
          {
            "considered_execution_plans": [
              {
                "plan_prefix": [
                ],
                "table": "`t1`",
                "best_access_path": {
                  "considered_access_paths": [
                    {
                      "access_type": "ref",
                      "index": "ix_b",
                      "rows": 2,
                      "cost": 2.4,
                      "chosen": true
                    },
                    {
                      "access_type": "range",
                      "range_details": {
                        "used_index": "ix_b"
                      },
                      "chosen": false,
                      "cause": "heuristic_index_cheaper"
                    }
                  ]
                },
                "condition_filtering_pct": 100,
                "rows_for_plan": 2,
                "cost_for_plan": 2.4,
                "chosen": true
              }

       --最终选择的扫描方式是ref,优化器认为使用range索引扫描方式成本会更高,heuristic_index_cheaper
            ]
          },
          {
            "attaching_conditions_to_tables": {
              "original_condition": "(`t1`.`b` = 2)",
              "attached_conditions_computation": [
              ],
              "attached_conditions_summary": [
                {
                  "table": "`t1`",
                  "attached": null
                }
              ]
            }
          },
          {
            "refine_plan": [
              {
                "table": "`t1`"
              }

         从以上的过程中,我们可以看出mysql对于一个sql语句主要分为三个部分(即使一个语句也称为join): join preparation,join

optimization,join exectuion.而optimizer对于join exectuion没有任何跟踪信息,optimzier只负责前两个部分

    在join preparation主要是对sql语句进行了一些初始化的处理,而且这个和mysql的版有一定的联系,mysql把我们发送过来的sql进行一定

的转换,再进行进一步的处理.

    在join optimization阶段主要进行了条件处理(相等传播,常量传播,条件化简),替代栏位,行评估,所有的计划分析,最后的执行选择等 .