ClickHouse查看执行计划(EXPLAIN语法)

1.EXPLAIN 语法示例

EXPLAIN [AST | SYNTAX | QUERY TREE | PLAN | PIPELINE | ESTIMATE | TABLE OVERRIDE] [setting = value, ...]

SELECT ... \| tableFunction(...) \[COLUMNS (...)\] \[ORDER BY ...\] \[PARTITION BY ...\] \[PRIMARY KEY\] \[SAMPLE BY ...\] \[TTL ...

]

FORMAT ...

查询计划的类型说明:

AST(抽象语法树):在AST级别优化之后的查询文本

SYNTAX(语法优化):在AST级别优化之后的查询文本,返回优化后的sql

QUERY TREE(查询树):在查询树级别优化之后的查询树

PLAN(执行计划):查询执行计划,默认是此值

PIPELINE(管道):查询执行管道

1.1AST(抽象语法树)

查询语法树执行计划

explain AST SELECT
path,code,v1
FROM
autorun_t_index
WHERE
index_id = 1227
AND cluster_id =27
AND code IN('queue_name', 'queue_allocatedmb', 'queue_availablemb','queue_allocatedvcores','queue_availablevcores')
AND update_date IN(SELECT MAX(update_date) FROM autorun_t_index WHERE index_id = 1227 AND cluster_id = 27 and update_date <='2023-08-04 01:38:48')

展示结果:

SelectQuery (children 3)
ExpressionList (children 3)
Identifier path
Identifier code
Identifier v1
TablesInSelectQuery (children 1)
TablesInSelectQueryElement (children 1)
TableExpression (children 1)
TableIdentifier autorun_t_index
Function and (children 1)
ExpressionList (children 4)
Function equals (children 1)
ExpressionList (children 2)
Identifier index_id
Literal UInt64_1227
Function equals (children 1)
ExpressionList (children 2)
Identifier cluster_id
Literal UInt64_27
Function in (children 1)
ExpressionList (children 2)
Identifier code
Literal Tuple_('queue_name', 'queue_allocatedmb', 'queue_availablemb', 'queue_allocatedvcores', 'queue_availablevcores')
Function in (children 1)
ExpressionList (children 2)
Identifier update_date
Subquery (children 1)
SelectWithUnionQuery (children 1)
ExpressionList (children 1)
SelectQuery (children 3)
ExpressionList (children 1)
Function MAX (children 1)
ExpressionList (children 1)
Identifier update_date
TablesInSelectQuery (children 1)
TablesInSelectQueryElement (children 1)
TableExpression (children 1)
TableIdentifier autorun_t_index
Function and (children 1)
ExpressionList (children 3)
Function equals (children 1)
ExpressionList (children 2)
Identifier index_id
Literal UInt64_1227
Function equals (children 1)
ExpressionList (children 2)
Identifier cluster_id
Literal UInt64_27
Function lessOrEquals (children 1)
ExpressionList (children 2)
Identifier update_date
Literal '2023-08-04 01:38:48'
Identifier TabSeparatedWithNamesAndTypes

1.2SYNTAX(语法优化)

语法优化前执行sql

explain SYNTAX SELECT
path,code,v1
FROM
autorun_t_index
WHERE
index_id = 1227
AND cluster_id =27
AND code IN('queue_name', 'queue_allocatedmb', 'queue_availablemb','queue_allocatedvcores','queue_availablevcores')
AND update_date IN(SELECT MAX(update_date) FROM autorun_t_index WHERE index_id = 1227 AND cluster_id = 27 and update_date <='2023-08-04 01:38:48')

语法优化后推荐的sql

SELECT
path,
code,
v1
FROM autorun_t_index
WHERE (index_id = 1227) AND (cluster_id = 27) AND (code IN ('queue_name', 'queue_allocatedmb', 'queue_availablemb', 'queue_allocatedvcores', 'queue_availablevcores')) AND (update_date IN ((
SELECT max(update_date)
FROM autorun_t_index
WHERE (index_id = 1227) AND (cluster_id = 27) AND (update_date <= '2023-08-04 01:38:48')
) AS _subquery132))

1.3QUERY TREE(查询树)

1.4PLAN(执行计划)

默认值就是PLAN,一般用来查看sql的执行计划,是否使用引擎,分区,索引等信息

未添加索引前

explain PLAN SELECT
path,code,v1
FROM
autorun_t_index
WHERE
index_id = 1227
AND cluster_id =27
AND code IN('queue_name', 'queue_allocatedmb', 'queue_availablemb','queue_allocatedvcores','queue_availablevcores')
AND update_date IN(SELECT MAX(update_date) FROM autorun_t_index WHERE index_id = 1227 AND cluster_id = 27 and update_date <='2023-08-04 01:38:48')

未添加索引查询计划

Expression (Projection)
CreatingSets (Create sets before main query execution)
Expression (Before ORDER BY)
Filter (WHERE)
SettingQuotaAndLimits (Set limits and quota after reading from storage)
ReadFromStorage (Log)
CreatingSet (Create set for subquery)
Expression ((Projection + Before ORDER BY))
Aggregating
Expression (Before GROUP BY)
Filter (WHERE)
SettingQuotaAndLimits (Set limits and quota after reading from storage)
ReadFromStorage (Log)

添加引擎,分区后优化结果

Expression ((Projection + Before ORDER BY))
SettingQuotaAndLimits (Set limits and quota after reading from storage)
ReadFromMergeTree

1.5PIPELINE(管道)

官网示例

EXPLAIN PIPELINE SELECT sum(number) FROM numbers_mt(100000) GROUP BY number % 4;

执行结果:

(Expression)
ExpressionTransform
(Aggregating)
Resize 4 → 1
AggregatingTransform × 4
(Expression)
ExpressionTransform × 4
(SettingQuotaAndLimits)
(ReadFromStorage)
NumbersMt × 4 0 → 1

相关推荐
偶像你挑的噻3 分钟前
3.Qt-基础布局以及事件
开发语言·数据库·qt
Dxy123931021618 分钟前
MySQL如何做读写分离架构
数据库·mysql·架构
毕设十刻1 小时前
基于Vue的考勤管理系统8n7j8(程序 + 源码 + 数据库 + 调试部署 + 开发环境配置),配套论文文档字数达万字以上,文末可获取,系统界面展示置于文末
前端·数据库·vue.js
合方圆~小文2 小时前
不同画面,三个镜头实时监控拍摄方案
数据结构·数据库·人工智能
ChrisitineTX3 小时前
凌晨突发Java并发问题:synchronized锁升级导致接口超时,排查过程全记录
java·数据库·oracle
极限实验室3 小时前
Easysearch 2.0.0 性能测试
数据库·性能优化
老华带你飞4 小时前
社团管理|基于Java社团管理系统(源码+数据库+文档)
java·数据库·vue.js·spring boot·后端
shayudiandian4 小时前
用LangChain打造你自己的智能问答系统
java·数据库·langchain
马克学长4 小时前
SSM特殊教育学校学生管理系统002k1(程序+源码+数据库+调试部署+开发环境)带论文文档1万字以上,文末可获取,系统界面在最后面
数据库·学生管理系统·特殊教育·ssm 框架
卿雪5 小时前
Redis 线程模型:Redis为什么这么快?Redis为什么引入多线程?
java·数据库·redis·sql·mysql·缓存·golang